DE69215773T2 - Adaptive synchronization device - Google Patents

Description

The present invention relates to communication systems, call traffic processing devices, methods for operating a communication system and methods for processing call traffic.

In digital telecommunications systems, it sometimes happens that subscriber communication equipment is clocked independently of the network communication equipment (e.g., the public switched telephone network) that interconnects the subscriber equipment. A particularly significant example is the CDMA (code-division multiplexed access) radiotelephone system, which is an important type of digital cellular mobile telephone system. In the CDMA system, network nodes containing radios, i.e., the mobile telephones and cell site base stations (cells), are synchronized with clock signals received by the cells from a GPS (global positioning system) satellite, while the radio switching systems that connect the base stations to each other and to the public switched telephone network by digital communications are synchronized with clock signals that can also be received from the GPS satellite, but are distributed through the telephone network.

For the purposes of this discussion, two series of events, signals, or operations are considered to be synchronized or synchronous with each other if (a) they either occur at the same nominal frequency or one occurs at a frequency that is an integer multiple of the frequency of the other, and (b) they occur in fixed phase relation to each other. Operations that are not synchronous are considered to be asynchronous for the purposes of this discussion.

By independently timing the operation of different units of a communication system, the assumption is destroyed that the units provide call traffic to each other at a predetermined constant and unchanging frequency at constant and unchanging times, ie with a fixed phase. Instead, the result of the independent Timing means that the units provide call traffic to each other at a rate and at times that fluctuate around a fixed frequency and phase. This asynchrony must be compensated for somehow.

Independent timing is only one cause of this asynchrony. Another cause that can exist in communication systems such as the CDMA radiotelephone system mentioned above is the lack of a predetermined and fixed propagation delay between the communicating units. Assuming that both the source unit and the destination unit are clocked either by a common clock or by different clocks that are synchronized with each other, if the propagation delay between the units is fixed and predeterminable, it can be compensated for in the communication system design so that the units can operate in synchronism with each other. However, if the delay cannot be predetermined but is variable and fluctuates, the net effect is the same as if the units were independently clocked. The variation in the delay can, for example, be a result of occasional changes in the transmission paths followed by communications between communicating entities, or variations in the volume of communication traffic that flows between and must be processed by the communicating entities. This asynchrony must also be compensated for.

A partial but inadequate solution to the problems caused by independent timing is to conduct communications between the communicating units in analog rather than digital form. Analog communications can be received asynchronously to their transmission. And although the asynchrony can introduce errors or noise into the communications, the problem is often tolerable in voice-only communications. For example, in the CDMA radiotelephone system, the radiotelephone switching systems can also be synchronized with the GPS satellite clock signals and therefore operate synchronously with the radio telephones and base stations when the switching systems are connected to the telephone network via analog voice-only communications. Of course, such an arrangement suffers from all the disadvantages associated with analog communications, such as low quality and capacity and susceptibility to interference, plus the problem of noise caused by asynchrony, which makes the arrangement unsuitable for data communications.

Similarly, a partial but inadequate solution to the problems caused by latency variation is circuit switching of communication traffic, which avoids the dependence of latency on the volume of communication traffic. However, circuit switching is ineffective or undesirable in many applications for other reasons. Furthermore, circuit switching does not eliminate the latency variation caused by path changes, such as typically occurs during a "soft handoff" of a CDMA connection.

According to one aspect of the present invention, a communication system according to claim 1 is provided.

According to a further aspect of the present invention, call traffic processing devices according to claim 9 are provided.

According to a further aspect of the present invention, a method according to claim 10 is provided.

According to yet another aspect of the present invention, a method according to claim 18 is provided.

In a telecommunications system which interconnects communicating units over a transmission medium which experiences varying transmission delay (such as a packet-switched communication medium), an interface is provided between the communicating units which is nominally synchronised with respective ones of the units, but which is adapted to operate within a predefined window, ie a range, of phase relationships with the other units and occasionally adjusts its otherwise fixed phase relationship to the operation of the respective units in order to achieve and maintain their operation within the predefined window.

The operations of the various units are thereby effectively synchronized with the operations of the interface arrangement and are therefore able to proceed essentially as if they were synchronized with one another and interconnected by a transmission medium with a fixed transmission delay.

More particularly, in a communications system comprising a first unit for either receiving incoming communications traffic or transmitting outgoing communications traffic at times dictated by first clock signals having a nominal frequency and a first phase, a second unit for either transmitting incoming communications traffic to the first unit or transmitting outgoing communications traffic received from the first unit at times dictated by clock signals having the nominal frequency, a third unit for interconnecting communications between the first and second units, and a communications medium connecting the second unit to the third unit for communicating communications traffic therebetween and having a varying transmission delay, the interconnection functions are performed as follows. The third unit either transmits incoming communications traffic received from the second unit to the first unit or transmits outgoing communications traffic received from the first unit to the second unit at times dictated by clock signals having the nominal frequency and having a second phase offset from the first phase by an adjustable fixed amount. A determination is made as to whether the second unit receives outgoing communication traffic from the third unit within predetermined time windows prior to the transmission times of the received outgoing communication traffic by the second unit, or whether the third unit receives incoming communication traffic from the second unit within predetermined time windows prior to the transmission times of the received incoming communication traffic by the third unit. Then, in response to a determination that receptions of either type of communication traffic fall outside the respective predetermined windows, the amount of offset of the second phase from the first phase is either increased or decreased as necessary to shift the subject receptions into the respective windows.

More particularly, according to an exemplary embodiment of the invention, in a CDMA radiotelephone system having radiotelephone switching systems and base stations (cells) interconnected by a packet transmission medium, each radiotelephone switching system includes a digital communications interface arrangement whose connections to the telephone network are synchronized with the operation of the telephone network and whose connections to the base stations (cells) are also nominally synchronized with the operation of the telephone system, but which operates for each individual connection within a predefined window of phase relationships to the operation of a base station handling the call and occasionally adjusts its phase relationship to the operation of the telephone system to compensate for transmission delay variations experienced by the medium to achieve and maintain its operation within the predefined window. Since the interface arrangement uses packet-switched communications between the base stations and the switching system, the phase relationship variations and timing adjustments are absorbed and hidden by variations in the spacing between packets. However, according to the exemplary embodiment, the interface arrangement uses circuit-switched communications between the switching system and the telephone network, wherein the timing adjustments are made by slips - bit insertions or deletions - in the communication bit stream. be absorbed.

The perceived asynchrony between the operations of the switching systems and the telephone network and the base stations (cells) resulting from transmission delay variations caused by variations in the amount of communication traffic carried by the medium or by changes in the transmission paths taken by the communication traffic during smooth handoff are compensated for and taken into account by the interface arrangement as required for the correct operation of the digital communication system.

While in the following discussion of an exemplary embodiment a distinction is made between Level 3 "packets" and Level 2 "frames", for the sake of clarity, the use of the term "packet" here and in the claims is intended to include either or both of the terms "packets" and "frames".

These and other advantages and features of the invention will become apparent from the following description of an exemplary embodiment of the invention in conjunction with the drawing.

Short description of the drawing

Figure 1 is a block diagram of a common cellular radiotelephone system;

Figure 2 is a block diagram of a cellular radiotelephone system incorporating an embodiment of the invention;

Figure 3 is a block diagram of a cell of the system of Figure 2;

Figure 4 is a block diagram of a cell interconnection module of the system of Figure 2;

Figure 5 is a block diagram of a speech coding module of the system of Figure 2;

Figure 6 is a block diagram of a speech processing unit of the module of Figure 5;

Figure 7 is a block diagram of a LAPD frame of the system of Figure 2;

Figure 8 is a block diagram of a modified LAPD frame of the system of Figure 2;

Figure 9 is a block diagram of a Level 3 protocol used to carry voice and/or signaling information in the frames of Figures 7 and 8;

Figure 10 is a block diagram of a Level 3 protocol used to carry signaling information in the frames of Figures 7 and 8;

Figures 11-14 are a flow diagram of receive packet processing functions of the processor of the unit of Figure 6;

Figure 15 is a flow diagram of transmit packet processing functions of the processor of the unit of Figure 6;

Figure 16 is a flow chart of clock setting functions of a multiple controller of the cell of Figure 3;

Figure 17 is a flow chart of clock adjustment functions of the processor of the unit of Figure 6 performed in step 970 of Figure 11;

Figure 18 is a flow chart of clock adjustment functions of the processor of the unit of Figure 6 performed in step 912 of Figure 11;

Figure 19 is a timing diagram of packet transmission timing adjustments performed during call setup for a service line of the unit of Figure 6;

Figure 20 is a timing diagram of packet reception timing adjustments performed during call setup for a service line of the unit of Figure 6;

Figure 21 is a timing diagram of packet transmission timing adjustments performed during an ongoing connection for a service line of the unit of Figure 6;

Figure 22 is a timing diagram of packet reception timing adjustments performed during an ongoing connection for a service line of the unit of Figure 6;

Figure 23 is a signaling diagram of the establishment of a mobile unit initiated call in the system of Figure 2;

Figure 24 is a signaling plan of the establishment of a network initiated connection in the system of Figure 2;

Figure 25 is a signaling diagram of a mobile unit initiated disconnection of a connection in the system of Figure 2;

Figure 26 is a signaling plan of a network-initiated disconnection of a link in the system of Figure 2;

Figure 27 is a signaling plan of the initiation of a soft handoff of a connection in the system of Figure 2;

Figure 28 is a signaling diagram of the end of a soft handoff with a pilot cell dropping;

Figure 29 is a signaling diagram of the end of a soft handoff with a slave cell dropping;

Figure 30 is a signaling diagram of a mobile unit initiated disconnection of a call during soft handoff in the system of Figure 2;

Figure 31 is a signaling plan of a network-initiated disconnection of a call during soft handoff in the system of Figure 2;

Figure 32 is a signaling plan of a semi-soft handoff of a connection in the system of Figure 2;

Figure 33 is a signaling plan of a hard CDMA-CDMA handoff of a call in the system of Figure 2;

Figure 34 is a signaling plan of a hard CDMA analog handoff of a call between cells served by the same cellular digital switch in the system of Figure 2; and

Figure 35 is a signaling plan of a hard CDMA analog handoff of a call between cells served by different cellular digital switches in the system of Figure 2.

Detailed description

Before beginning a description of an embodiment of the invention, it may be useful to consider an existing cellular mobile telephone system to serve as a basis for comparison. One such system is shown in Figure 1. A description of such a system can be found in K. W. Strom, "On the Road with AUTOPLEX System 1000," AT&T Technology, Vol. 3, No. 3, 1988, pages 42-51 and W. J. Hardy and R. A. Lemp, "New AUTOPLEX Cell Site Paves The Way For Digital Cellular Communications," AT&T Technology, Vol. 5, No. 47, 1990, pages 20-25.

The system of Figure 1 includes a plurality of geographically distributed service nodes known as cell sites or, for short, cells 102, each providing radiotelephone service to wireless user terminals known as cellular telephone sets 103 in their vicinity. To provide radiotelephone service between mobile radiotelephone sets 103 served by different cells 102 and between mobile radiotelephone sets 103 and the public telephone network 100, the cells 102 are connected to one another and to the network 100 via cellular telephone switching nodes referred to herein as digital cellular switches (DCS) 101. Each switch 101 is, for example, the cellular digital switch of AT&T's Autoplex cellular telecommunications system. Each cellular digital switch 101 is connected to a plurality of different cells 102 via communication lines 107 and is connected to the network 100 via communication lines 106. Each line 106 and 107 is, for example, a DSO (64 kB/s time division multiplex) channel, a majority of which are implemented by a DS1 device that can be transmitted via land line (T1 line), optical transmission, microwave, etc. devices. Control of the system of Figure 1 and coordination of the activities of the various cells 102 and DCS 101 is performed by an Executive Cellular Processor (ECP) 105 which is connected to each cell 102 and cellular switch 101 via an Interprocess-Message Switch (IMS) 104 through control links 108. Together, ECP 105 and IMS 104 form an ECP complex 134. ECP complex 134 and DCS 101 form a mobile switching center (MSC) 199. ECP 105 and IMS 104 are, for example, AT&T's Autoplex ECP and AT&T's Autoplex IMS (which includes a plurality of cell site node processors, digital switching node processors, and database node processors interconnected via an IMS ring), and links 108 are, for example, RS-449 data links within MSC 199. Alternatively, control links 108 may be implemented as 64 kB/s DS0 channels on DS1 facilities between cells 102 and mobile switching center 199.

Each cellular telephone 103 typically includes an analog FM radio telephone that is operable at any of a plurality of radio frequency pairs. Each cell 102 includes a plurality of analog FM radios 143, each operating at one of the radio frequency pairs of the cellular telephones 103. Radios 143 of neighboring cells 102 operate at different frequency pairs to avoid mutual interference. However, each cellular telephone 103 is typically capable of operating at any of the frequency pairs of all cells 102.

In an alternative embodiment, the analog FM radios and radiotelephones are replaced by digital radios and radiotelephones operating with time-division multiple-access (TDMA). In this embodiment, vocoder functions can be added to the radio units or be located at switches 101. While located at a cellular system, a cellular telephone receiver polls a set of predetermined paging channels. After locking onto the strongest paging channel, cellular telephone 103 receives instructions from the system and receives incoming calls. A cellular telephone 103 also transmits on a channel to initiate a call. Once a call is made (incoming or outgoing), the receiver is assigned a particular voice channel and instructed to tune to that transmit and receive frequency pair. At the same time, a connection is established between cell 102 and telephone network 100 via a cellular digital switch 101, which provides the voice path for the long distance call.

Once this voice connection has been established, the radio signal levels are monitored by the cell's radio 143. As the mobile radiotelephone 103 moves from one cell to another, the serving cell 102 detects the decrease in signal strength and requests that measurements be taken from surrounding cells 102. If these measurements indicate that another cell 102 can provide better service, then the voice connection is switched to that cell 102 via a process known as "hard handoff." The hard handoff process is under the control of ECP 105 and requires that a DCS 101 first establish a 3-wire connection that extends the voice path from the serving line 106 to radio channels in both the serving cell 102 and the target cell 102. Upon confirmation of this connection, the radiotelephone 103 is instructed to retune to the frequency of the assigned radio 143 in the target cell 102. After confirming the radiotelephone's communication with the target cell 102, the DCS 101 is then instructed to remove the voice connection with the original serving cell 102 and to establish the connection between the new serving (target) cell 102 and the serving line 106. During this handoff process, the long distance call continues largely uninterrupted. In the meantime, the original voice channel is made available for use by another subscriber. Hard handoffs performed in this manner utilize processor capacity in both the ECP complex 134 and the cellular digital switch 101. For the duration of the three-way connection, the hard handoff also utilizes additional switching fabric capacity (the TDM bus 130). If the target cell 102 containing the selected radio 143 is connected to a switching module 120 other than that containing the serving line 106, then the connection must be continued over a time-multiplexed switch (TMS) 121 using additional switching fabric in that switching element. As the number of cells 102 in a system increases, the number of handoffs increases, using an increasing proportion of the system processor and switch fabric resources, and thereby reducing the overall capacity of the system.

Each cell 102 is constructed around a high rate time-division multiplexed (TDM) bus 140. The TDM bus 140 is, for example, the 2.048 MHz TDM bus of an AT&T Definity universal communication system module and physically comprises one or more TDM buses, each with 256 time slots per frame. For example, multiple TDM buses are used simultaneously by devices connected thereto and logically function as a single TDM bus with a plurality of 256 time slots per frame. Each time slot has a rate of 64 kB/s. Within a cell 102, radios 143 are connected to the TDM bus 140. Radios 143 receive communications for radio transmission and deliver received radio communications to the TDM bus 140 in DS0 channel format at a rate of 64 kB/s. The input to and output from each radio is pulse code modulation (PCM) encoded speech with full bit rate. Also connected to the TDM bus 140 are one or more interfaces 142, each of which couples the TDM bus 140 to lines 107. For example, lines 107 are passed through T1 devices using the DS1 communications format, which operate at a bit rate of 1.544 Mb/s, and so the interfaces 142 are DS1 interfaces. The DS1 and DS0 formats mentioned above are described by T.H. Murray in "The Evolution of DDS-Networks: Part 1," Telecommunications, February 1989, pages 39-47. An interface 142 accepts communications from the TDM bus 140 supplied by a plurality of radios 143, multiplexes them into the DS1 format, and transmits them on lines 107. In the reverse direction, interface 142 receives communications formatted in DS1 format from lines 107, de-interleaves them, and delivers them to TDM bus 140 for transmission to radios 143. TDM bus 140 operates under the control of a controller 141 which assigns time slots on bus 140 to individual ones of radios 143 and interfaces 142. For example, controller 141 makes these assignments based on control information supplied from ECP complex 134 via control link 108; alternatively, controller 141 may have a database which allows it to make the assignments automatically.

Each cellular digital switch 101 includes one or more digital switching modules (DSM) 120. A module 120 is similar in structure to a cell 102 in that it includes a TDM bus 130 similar to the TDM bus 140, a controller 131 that provides the same TDM bus control functions as the controller 141, and a plurality of interfaces 132 connected to the bus 130 that provide the same functions as the interfaces 142. Based on control communications initiated by the ECP complex 134, the controller 131 causes communications to be switched between interfaces 132 through the TDM bus. 130. Each line 107 extending from a cell 102 is terminated by an interface 132 on a switching module 120. Lines 106 which are copies of lines 107 but extend to the public switched telephone network 100 are terminated by other interfaces 132 on a module 120.

If the switch 101 contains more than one module 120, it also contains a time multiplexed switch (TMS) 121. A TMS interface 133 is then connected to the TDM bus 130 in each module 120 and terminates a link 109 that extends to the TMS 121. The interface 133 is, for example, the Module Control Complex (MCC) of an AT&T Definity universal communication system module. The TMS 121 provides direct-switched interconnection between modules 120 of a cellular telephone switch 101. Interconnection between modules 120 of different cellular telephone switches 101 is provided by the public switched telephone network 100 or by lines that directly interconnect the switches 101.

Overall control of a cellular digital switch 101 and coordination of activities between its modules 120 and 121 is performed by a DCS controller 161. The DCS controller 161 is in direct communication with the ECP complex 134 via a control link 108. The controller 161 has its own control connection to the TMS 121 via link 150 and to controllers 131 of switch modules 120 via link 150 and TMS interfaces 133. The controller 161 is, for example, the processor 501 CC of an AT&T Definity communications system.

Turning now to Figure 2, it shows an embodiment of a cellular mobile radio telephone system constructed in accordance with the invention. The same numerical designations as in Figure 1 are used in Figure 2 to designate elements that are common to both systems.

Figure 2 shows a system topology that corresponds to 1, although not identical. The system of FIG. 2 includes a plurality of geographically dispersed cells 202, each providing radiotelephone service to cellular telephones 203 in its vicinity. As used herein, cell 202 refers to either a geographically separate cell site or to one of a plurality of "sites" at a given cell site, where a "site" is a cell sector as typically implemented by the use of transmit directional antennas at a cell site. The operation of all cellular telephones 203 and cells 202 is synchronized to a common master clock, such as clock signals generated and broadcast by a global navigation system satellite. Interconnection between cells 202 and between cells 202 and the public switched telephone network 100 is accomplished by cellular digital switches 201 in two stages. First, individual cells 202 are connected by lines 207 to one or more cell interconnect modules (CIM) 209 of a DCS 201. Cell interconnect modules 209 of individual DCS 201 are in turn connected by fiber optic packet switching lines 210 to each speech coding module (SCM) 220 of that DCS 201. The cellular digital switches 201 are each connected to the public network 100 by a plurality of lines 106, analogous to Figure 1, and directly to one another via lines 206 which are functional copies of the lines 106. The operation of the switches 201 is synchronized with master clock signals (not shown) of the public telephone network 100. Still analogous to Figure 1, cells 202 and cellular digital exchanges 201 are operated under the control of the ECP complex 134, to which they are connected via control links 108. Likewise, the various modules 209 and 220 of a DCS 201 are connected via control links 208 to a common DCS controller 261 and are operated under their control. Physically, the DCS controller 261 is, for example, again the CC processor 501.

In the system of Figure 2, some, but not necessarily all, of the cellular telephones 203 are digital radio telephones. Although shown as being mounted in a vehicle for illustrative purposes, a cellular telephone 203 may be any portable radio telephone and may even be a stationary radio telephone. The digital radio telephones use voice compression techniques to reduce the required digital transmission rate over the radio channel. Each digital radio telephone includes voice compression circuitry in its transmitter and voice decompression circuitry in its receiver. Each radio telephone is capable of operating on any of a plurality of wideband radio frequency pairs.

To handle non-packetized traffic analogous to that handled by the system of Figure 1 side-by-side with packetized traffic, a DCS 201 of the system of Figure 2 includes the elements shown in dashed lines: a TMS 121 connected by lines 109 to modules 209 and 220, and lines 106 connecting CIM 209 directly to the public switched telephone network 100. Their use is discussed below.

Digital radiotelephones 203 may operate in one or more of time-division multiple-access (TDMA) or code-division multiple-access (CDMA) modes, or any other digital or analog radio mode. TDMA is a technique known in the art that provides multiple users with access to a radio channel (frequency) by dividing that channel into multiple time slots. An individual user may be assigned to one or more of these time slots. An example of a TDMA radio 203 is the TIA IS54 cellular digital radio. In TDMA, different frequencies are assigned to neighboring cells and therefore requires the "hard switching" procedure already described.

In the present embodiment, it is assumed that digital radiotelephones 203 are operating in the CDMA mode or, failing that, in the (analog) FDMA mode. CDMA is a DSSS (direct sequence spread spectrum) technique in which the frequencies in the areas served by neighboring cells 202 can be reused. As a result, neighboring cells 202 do not need to operate on different radio frequencies and will not, but will reuse the same frequencies. When a mobile telephone 203 moves from the vicinity of one cell 202 to the vicinity of another cell 202, it may be subject to a "hard handoff" procedure as previously described. However, a CDMA mobile telephone 203 in the system of Figure 2 may alternatively and preferably be subject to a "soft handoff" procedure during which it communicates with both cells 202 on the same frequency pair at the same time. The CDMA technique and its associated methods and devices are also well known in the art. The basic principle of direct-sequence code division multiple-access is to use a plurality of individual and separate high-rate digital signals that are absolutely or statistically orthogonal to each other to modulate one of a plurality of low-rate (i.e., baseband) user signals and to combine the plurality of modulated signals into common digital signals that are then used to control high-frequency modulation functions. Recovery and separation of the original baseband signals is achieved by using the corresponding digital modulation signals for demodulating in a time-synchronous manner. A description of CDMA can be found in, for example, US Patent No. 4,901,307 and published international patent applications WO91/07020, WO91/07036 and WO91/07037.

A cell 202 is shown in Figure 3. Like a cell 102 of Figure 1, the cell 202 includes a TDM bus 140 operating under control of the controller 241, and DS1 interfaces 242 couple the TDM bus 140 to lines 207. The controller 241 is, for example, the control complex of an AT&T Autoplex Series II cell site. Functionally, it is a copy of the controller 141 of a cell 102, but now performs additional functions described below since the cell 202 includes a plurality of digital radios 243. The signal input and output of each digital radio are coupled to the TDM bus 140 through one or more corresponding channel elements 245 and a multiplex controller 244. A channel element 245 is an interface to digital radios 243 serving a single user. Channel elements 245 provide signal processing functions - in this example, baseband and spread spectrum (CDMA) signal processing functions - for individual calls transmitted and received by their associated radios 243.

Each multiple controller 244 includes a C-bus 390. The C-bus 390 is, for example, a conventional computer input and output (I/O) bus, and the channel elements 245 are connected to the C-bus 390 as computer I/O devices. The C-bus 390 and channel elements 245 operate under the control of a controller 393. The controller 393 is, for example, a general purpose microprocessor and is served by a bus 391, which is, for example, a conventional microprocessor main bus. The bus 391 is connected to the C-bus 390 by a C-bus interface 392 which functions as an I/O interface of conventional design. The controller 393 orchestrates data movement between channel elements 245 and the TDM bus 140 of the cell 202 (e.g., one transfer in each direction for each channel element 245 every 20 msec), performs operations, management, and maintenance (OAM) functions at the multiple controller 244, handles cell location signaling and other specialized functions, and performs formatting and Level 2 and Level 3 protocol deformatting functions are performed on data (call traffic and signaling) passing between channel elements 245 and TDM bus 140. A memory 394 is connected to bus 391 and serves as a scratch pad traffic buffer and command memory for controller 393. Also connected to bus 391 is an HDLC controller 395. It performs HDLC formatting and deformatting functions on traffic passing between channel elements 245 and TDM bus 140, including traffic conversion between the byte-oriented form used in multiple controller 244 and the bit-oriented form used on TDM bus 140, including bit stuffing and LAPD flag insertion functions. The HDLC controller 395 receives and transmits HDLC serial bit streams from/to the TDM bus 140 via a TDM bus interface 396 of conventional design that connects the controller 395 to the bus 140.

Compressed call traffic and signaling are transported in the form of segments of byte-oriented information between channel elements 245 and the multi-controller 244. Each channel element 245 transmits and receives a segment of byte-oriented information at regular intervals, for example, every 20 msec. The multi-controller 244 formats each segment of byte-oriented information into the LAPD protocol format, which includes a Level 3 protocol, for transmission to the DCSs 201. Although any suitable Level 3 protocol may be used, example Level 3 protocols 350 and 351 are shown in Figures 9 and 10.

Figure 9 shows a protocol 350 used to convey either call traffic or signaling or both, while Figure 10 shows a protocol 351 dedicated to conveying a particular type of signaling. The two protocols 350 and 351 are carried by frames of Figures 7 and 8. A data element of the Level 3 protocol carried by a Level 2 protocol is usually referred to as a packet and a data element of the Level 2 protocol is commonly referred to as a frame. Protocol 350 of Figure 9 includes at least information fields 320-327. Additional fields for other types of information may be included in packet 350 but are not relevant to the present description. Sequence number field 320 carries a sequence number of this packet 350 within the packet sequence transmitted in a given direction. In the case of packet 350 outgoing to a channel element 245 from a DCS 201, the sequence numbers start with zero at the beginning of each new call. In the case of packets 350 incoming from a channel element 245 to a DCS 201, the sequence numbers are derived from the master clock signals to which all mobile telephones 203 and cells 202 are synchronized. In the packet type field 321, the packet type is identified as either a traffic packet corresponding to packet 350 of Figure 9 or a signaling packet corresponding to packet 351 of Figure 10. In the clock adjustment field 322, information is passed from the multiple controllers 244 to the DCSs 201 that is used to correct real and virtual drift between the master clock to which mobile telephones 203 and cells 202 are synchronized and a master clock to which the public switched telephone network 100 and the DCSs 201 are synchronized. The field 322 is used only in the reverse direction and is zero in the forward direction. The radio transmission CRC field 323 contains the result of a common checksum calculated by a mobile telephone 203 over its transmitted traffic, which is sent along with that traffic by the mobile telephone 203. The signal quality field 324 carries records calculated by channel elements 245 of the quality of call traffic signals received from the mobile telephone 203. Fields 323 and 324 are also used only in the reverse direction and are zero in the forward direction. The power control field 325 carries information from a cell 202 regarding the tendency of power transmitted by a channel element 245 to its power control commands sent to the respective mobile telephone 203. Normally, this field is also used only in the reverse direction, but is used in both directions during a soft handoff, as will be explained. The voice/signaling type field 326 identifies the type of information carried by the packet 350: voice traffic only, voice plus signaling, or signaling only. And the voice/signaling data field 327 carries connecting voice traffic or signaling information, or a mixture of both, to and from the channel elements 245.

A signaling packet 351 shown in Figure 10 is simpler than the traffic packet 350 of Figure 9: it has fields 321 and 328-331 which are relevant to this discussion. The packet type field 321, already discussed in connection with Figure 9, identifies the packet 351 as a signaling packet. The message type field 328 identifies the type of signaling carried by the packet 351. The channel element identification field 329 identifies the particular channel element 245 participating in this message exchange. The frame selector identification field 330 identifies a particular virtual port on a processor 602 (see Figure 6) participating in this message exchange. These fields 329 and 330 may be used for security, maintenance, performance tracking, billing, routing, etc. purposes. Channel element 245 and frame selector identifiers are administratively assigned at system configuration time and remain fixed thereafter. And signaling data field 331 maintains the transmitted signaling information.

A plurality of channel elements 245 are coupled to the TDM bus 140 by a multiple control unit 244. Each multiple control unit 244 communicates on the TDM bus 140 via an assigned input and output "line". The assignment can be managed and typically occurs at system initialization. Each "line" consists of, for example, a plurality of (eg four) time slots (ie four 64-kB/s channels) on TDM bus 140. In the reverse (incoming) direction, traffic segments received from channel elements 245 are queued by multi-controller 244, formatted into packets, packetized into LAPD (Layer 2 Protocol) frames in inverted HDLC format, and transmitted one at a time into their associated output "line" on TDM bus 140. In the forward (outgoing) direction, multi-controller 244 receives LAPD frames from its associated input "line" on TDM bus 140, completes the LAPD protocol, deformats the packets, and then distributes the contents of those packets to channel elements 245 according to an address field embedded in the receive frame. As a result of the operations of the multiple controllers 244, frames transmitted to and from them are statistically multiplexed onto the TDM bus 140, thereby greatly increasing the bandwidth performance of the TDM bus 140 over alternative transmission methods.

An exemplary LAPD frame 300 is shown in Figure 7. For purposes of this discussion, it includes a plurality of fields 301-305; a flag field 301 used to delimit the frame; a data link connection identifier (DLCI) field 302; a control field 303 which indicates what type of LAPD frame this is; a payload field 304 which contains the above-mentioned layer 3 protocol (packet) 350 or 351; and a frame check sequence (FCS) field 305 which is used for error checking. The DLCI field 302 is the end-to-end address field of the frame. It contains a number or index (DLCI) of the virtual connection which associates the frame with a particular call. In the forward direction, the DLCI identifies a specific channel element 245, in the reverse direction, the DLCI identifies a specific one of a plurality (for example two) of virtual ports of the processor 602, which correspond to a specific service line 612 of the speech processing unit 264 (see Figure 6). Within a Multiple control unit 244, the DLCI identifies the channel element 245 that is the source or destination of the frame. In the present embodiment, DLCI are administratively assigned to ports and channel elements at system configuration time and remain fixed thereafter.

The transmission of frames to and from the multiple control units 244 is accomplished using the frame forwarding transmission technique, whereby protocol termination of the frames occurs only at the transmission endpoints, thus greatly increasing the efficiency and speed of these frame transmissions through the system of Figure 2. The frame forwarding technique is described in U.S. Patent No. 4,894,822. It is hereby incorporated by reference herein.

Advantageously, to provide radiotelephone services to conventional analog or digital TDMA mobile telephones 103 within the same system, analog FM or digital TDMA radios 143 may also be connected to TDM bus 140 in cells 202 in the manner described for cells 102, as indicated by the dashed blocks in Figure 3. Alternatively, conventional cells 102 may be used side by side with cells 202 within the system of Figure 2. TDMA traffic may be carried through the system of Figure 2 either in circuit-switched form like analog radio traffic or in packet-switched form like CDMA traffic.

In cell 202 of Figure 3, DS1 interfaces 242 perform their conventional functions of collecting 64 kB/s time slots from TDM bus 140 and multiplexing them into DS1 format for transmission on lines 207 and vice versa. For the purposes of this application, it is important that each interface 242 ensure that the delay of signals of each DS0 channel within interface 242 is constant; many commercially available DS1 interfaces, such as AT&T's TN 464C, also meet this requirement.

Due to the functions performed by the multiple control units 244, frames are statistically multiplexed onto lines 207 and the format of devices implementing the lines 207 is no longer, from a logical perspective, the purely conventional DS1 format of devices implementing the lines 107 of Figure 1: instead of comprising 24 independent DS0 channels as in the case of DS1 devices, each device now comprises several independent "lines", each consisting of the bandwidth of one or more DS0 channels. Each of the "lines" carries the LAPD frames created by or intended for a single multiple control unit 244. The traffic performance of the bandwidth provided by the lines 207 is thereby greatly increased over alternative transmission methods such as the conventional circuit switching method. All remaining lines 207 (i.e., DS0 channels) that are not bundled into "lines" continue to be used on an independent, single circuit-switched basis, for example, to transport communications to and from conventional radios 143.

A cell interconnect module (CIM) 209 is shown in Figure 4. The cell interconnect module 209 is based, for example, on the universal module of AT&T's Definity communications system. It includes a local area network (LAN) bus 250 which operates under the control of a controller 251. Universal DS1 (UDS1) interfaces 252 connect the lines 207 to the LAN bus 250. Each interface 252 includes a DS1 line interface 442 which copies the DS1 device interface circuit of the DS1 interface 242 and a packet processing element (PPE) 401 which are interconnected by a bus 400. The bus 400 is a time division multiplexed bus with 64 time slots each having a bit rate of 64 kB/s. The DS1 line interface 442 performs the functions of collecting 64 kB/s time slots from the collection bus 400, inverting the (in conjunction with the Cell 202 of Figure 3 discussed) back to its normal state and multiplexing the data into the DS1 format for transmission on lines 207 and vice versa.

The PPE 401 performs LAPD frame forwarding functions between the bus 400 and LAN bus 250. PPE 401 contains a translation table 411 that contains a card and a port address for each link layer address 302. The translation table 411 is maintained at initialization. PPE 401 is maintained for receiving LAPD frames 300 at assigned time slots of the bus 400. For each LAPD frame 300 received on the bus 400, the PPE 401 uses the contents of the frame's DLCI field 302 to find the corresponding card and port address in the table 411. The card and port addresses identify the destination of frame 300 on LAN bus 250. PPE 401 then extracts flag field 301 from frame 300 and prepends the found card and port addresses to the frame to form a modified LAPD frame 310 shown in Figure 8. Comparison with Figure 7 shows that flag field 301 has been replaced by card address 311 and port address 312. PPE 401 then transmits modified LAPD frame 310 on LAN bus 250. In the other direction, PPE 401 examines modified LAPD frames 310 transmitted on LAN bus 250 for its map address 311. It receives each frame 310 with the sought-after address 311, strips addresses 311 and 312 from frame 310, replaces them with flag field 301 to form LAPD frame 300, and then transmits frame 300 on bus 400. The stripped port address 312 identifies to PPE 401 the specific time slots at which that specific frame 300 is to be transmitted.

Expansion interfaces (EI) 253 are also connected to the LAN bus 250 of the cell interconnection module 209. A fiber optic cable 210 is routed through each expansion interface 253. coupled to the LAN bus 250. The expansion interfaces 253 act only as directional elements. Each expansion interface 253 contains a LAN bus interface 450 which monitors the LAN bus for modified LAPD frames 310 with a prefixed DLCI 302, card address 311 and port address 312. Interface 450 captures all frames 310 with the prefixed DLCI 302, card address 311 and port address 312, blanks out the prefixed card address 311 and stores the frame 310 in a FIFO buffer 451. From the FIFO buffer 451, the prefixed port address 312 and DLCI 302 of the frame 310 are output to a conversion table 452 and the fields 302-305 of the frame 310 are output to a conversion inserter 453. Table 452 is a presented table of card and port addresses of the speech encoder modules 220. The table 452 uses the port address 312 and DLCI 302 it receives from the FIFO buffer 451 as pointers to find a new card address 311 and port address 312 for the frame 310 and sends the new addresses 311 and 312 to the translation inserter 453. The inserter 453 prepends the new card and port addresses 311 and 312 received from the table 452 to the fields of the frame 310 it receives from the FIFO buffer 451 and sends the new frame 310 to the fiber interface 454. If no corresponding addresses are found in the table 452 and sent out by it, the inserter 453 simply discards the received frame 310. The fiber interface 454 transmits the frame 310 on the fiber optic line 210. Any desired protocol and transmission format can be used on the lines 210. In the reverse direction, the fiber interface 454 receives frames 310 on the line 210 and stores them in a FIFO buffer 455. The stored frames 310 are read out of the FIFO buffer 455 by the LAN bus interface 450 and transmitted on the LAN bus 250. The expansion interface 253 consequently transmits on the LAN bus 250 only those frames 310 that it received on the connected fiber line. 210. These frames 310 have card addresses 311 that identify the destination interfaces 252 on the LAN bus 250 and port addresses 312 that are not sought by any expansion interfaces 253 on the LAN bus 250.

For purposes of handling conventional circuit-switched cellular radiotelephone communications, cell interconnect module 209 includes the elements shown in dashed lines in Figure 4. In particular, CIM 209 includes a TDM bus 230 corresponding to TDM bus 130, and each UDS1 interface 252 includes a time-slot interchanger (TSI) 402 coupling the collector bus 400 to the TDM bus 230. TSI 402 performs the usual time-slot interchange functions. It receives dedicated 64 kB/s channels (time slots) on the bus 400 and TDM bus 230 and transmits them on dedicated time slots on the TDM bus 230 and bus 400, respectively. TSI 402 is programmed on a call-by-call basis. For the purpose of switching these conventional communications, the TDM bus 230 is coupled to a TMS 121 via a TMS interface 133 and line 109 in the manner described for Figure 1 (see Figure 2). For the purpose of connecting these conventional communications to the public switched telephone network 100, the TDM bus 230 is also coupled to the network 100 via a DS1 interface 132 and line 106.

Figure 5 shows a speech encoder module 220 of a cellular digital switch 201. Each DCS 201 comprises one or more identical modules 220. Module 220 is, for example, the universal module of AT&T's Definity communications system. Module 220 includes TDM bus 130 and a LAN bus 260, which is a copy of LAN bus 250, both of which operate under the control of a controller 231. As in Figure 1, TDM bus 130 is connected to public switched telephone network 100 via DS1 interfaces 132 and lines 106. Fiber lines 210 from the cell interconnect modules 209 are connected to the LAN bus 260 via expansion interfaces 263, which are copies of the expansion interfaces 253. Each cell interface module 209 of a DCS 201 is connected to each speech encoder module 220 of that DCS 201. Interconnection between DCSs 201 is provided by the network 100 via lines 106.

Buses 260 and 130 are interconnected by a plurality of call processing nodes referred to herein as voice processing units (SPU) 264. Based on the map address 311 prepended to each frame 310 by the expansion interfaces 253 of the cell interconnect modules 209, each voice processing unit 264 receives frames 310 addressed to it, depackets their contents (i.e., completes its protocol), performs various processing functions -- including voice decompression -- on the contents of each received frame, and outputs the processed frame contents on the TDM bus 130 in time slots assigned to the call-based calls. In the reverse direction, a voice processing unit 264 receives communications over the TDM bus 130 at time slots allocated to connections on a per-call basis, performs various processing functions - including voice compression - on them, packetizes the processed communications, inserts into each frame a DLCI 302 identifying a particular channel element 245 of a particular cell 202, prepends each frame with card and port addresses 311 and 312 identifying the frame destination on the LAN bus 260, and transmits the frames 310 on the LAN bus 360.

As a result of the operations of cell interconnection modules 209 and speech encoder modules 220, the frames 310 transmitted between them are statistically multiplexed onto lines 210 and frame-forwarded thereover, thereby greatly increasing the traffic performance of the bandwidth provided by lines 210 compared to alternative transmission methods such as circuit switching.

As already mentioned in connection with Figure 3, the DCS 201 optionally contains a TMS 121 for operating conventional radiotelephone communications. The speech encoder module 220 is connected to TMS 121 via a line 109 and a TMS interface 133 in the manner described for the switching modules 120 of Figure 1.

An example of a speech processing unit 264 is shown in Figure 6. Each SPU 264 includes a LAN bus interface 601. It monitors frames 310 passing through the LAN bus 260 for pre-allocated card addresses 311 and captures those with the addresses sought 311. The LAN bus interface 601 includes a buffer 620. When a frame 310 is captured, the LAN bus interface 601 appends a timestamp to it, stores it in the buffer 620, and issues an interrupt to a processor 602. The timestamp is the current count of a counter 623, discussed below.

The port address 312 of a frame 310 identifies one of a plurality of service lines 612 implemented by the SPU 264. A service line 612 is assigned to a connection either for the duration of the call or until a hard handoff occurs. Each service line 612 has its own tone processing circuitry. However, all service lines 612 are serviced on a time-sharing basis by the processor 602, which performs frame selection and protocol processing functions for all service lines 612 of an SPU 264. The functions performed by the processor 602 on frames 310 received from the LAN bus interface 601 are shown in Figures 11-14 and 17-18, and functions performed by the processor 602 on traffic segments received from service lines 612 (hereinafter also referred to as traffic frames) are shown in Figure 15. The processor 602 performs each of these functions for each service line 612 every 20 msec. The execution of the functions is interrupt-controlled by interrupt signals provided by an adaptive synchronization circuit 611 and interface 601.

The exchange of traffic frames of incoming and outgoing call traffic is carried out between the processor 602 and the service lines 612 via buffers 603 of the processor 602. Each service line 612 has its own corresponding buffer 603. In a buffer 603, traffic frames passing between the processor 602 and a vocoder 604 of a service line 612 are temporarily stored to compensate for minor differences and variations in the timing of input and output operations of the processor 602 and vocoder 604.

Each service line 612 has its own vocoder 604. Vocoders 604 provide speech compression and decompression functions. Each is a digital signal processor that receives a traffic frame of compressed speech from processor 602 via buffer 603 at regular intervals (e.g., every 20 msec) and decompresses the traffic frame into a predetermined number (e.g., 160 bytes) of pulse code modulated (PCM) speech samples. In the present example, each byte has a duration of 125 msec, referred to as a "tick." In the opposite direction, a vocoder 604 receives 160 bytes of PCM speech samples, performs speech compression functions on them, and outputs a traffic frame of compressed speech to processor 602 via buffer 603 at regular intervals (every 20 msec). Traffic frame exchange operations between vocoder 604 and processor 602 are clocked with clock signals generated by internal input and output clocks 621 and 622 of vocoder 604, while reception and transmission of PCM samples by vocoder 604 are clocked with clock signals generated by clock circuit 600. Clocks 621 and 622 are edge synchronized with clock signals from circuit 600 upon system initialization and service line 612 reset. Vocoders are well known in the art. Each vocoder 604 is configured, for example, using digital signal processor (DSP) 16A by AT&T, which implements the low-speed, variable-bit-rate speech coding/decoding algorithm QCELP from Qualcomm. Inc. The QCELP algorithm enables minimal information to be sent during periods of little or no speech activity. The frame transport mechanism of the present embodiment adapts ideally to time-varying traffic loads.

In the case where a system handles both CDMA and TDMA traffic, with the TDMA traffic also being frame forwarded, some of the service lines 612 are dedicated to handling the TDMA traffic and their vocoders 604 are, for example, AT&T's digital signal processor 16A programmed according to the TIA IS-54 standard for TDMA communications.

On their way from the vocoders 604, PCM samples pass through tone fader circuits 605. Each service line 612 has its own tone fader circuit 605. When commanded by processor 602, tone fader circuit 605 momentarily blocks and discards PCM samples output by vocoder 604 and replaces them with PCM samples of the respective tone selection signals specified by the command. Tone fader circuit 605 has no influence on the input of the PCM samples to vocoder 604. Operation of tone fader circuit 605 is synchronized with the output of vocoder 604 by clock signals generated by clock circuit 600.

Echo cancellers 606 are connected downstream of tone fader circuits 605 in the sequence of circuits of service line 612. Each service line 612 has its own echo canceller 606. Each cancels echoes of call traffic destined for telephone network 100 from call traffic originating from telephone network 100 by retaining an attenuated duplicate of the network-destination traffic generated by the vocoder and subtracting a correspondingly delayed duplicate from the received network-destination traffic. Echo cancellers are well known in the art. The timing of the operations of the Echo canceller 606 is controlled by clock signals generated by clock circuit 600.

Echo cancellers 606 receive network-originated traffic from and transmit network-destined traffic to a master bus 607. Master bus 607 is a passive TDM serial bus carrying 64 kB/s time slots. Each echo canceller 606 is statically assigned its own input time slot and its own output time slot on master bus 607.

The bus 607 is coupled to the TDM bus 130 via a TDM bus interface 608. The interface 608 performs time-slot interchange (TSI) functions between the bus 607 and the bus 130. Its operation is clocked by clock signals generated by the circuit 600 and is controlled by a translation and maintenance (XLATION AND MTCE) unit 609. The unit 609 performs time slot assignment functions from the bus 607 to the bus 130 on a call basis under the direction of the controller 231 of this speech encoder module 220. The unit 609 communicates with the controller 231 via a control channel implemented by the bus 130. This control channel is coupled to the unit 609 via the interface 608 and the bus 613. The unit 609 provides maintenance functions for the LAN bus interface 601 via the control path 616.

Unit 609 exercises control over interface 608 through a conversion and maintenance control bus 613 to which both are connected. Similarly, processor 602 controls circuits 601, 603-606 and 611 through a processor control bus 610. Communication between processor 602 and unit 609 is facilitated by a buffer 614 which couples bus 610 to bus 613.

The clock circuit 600 is connected to the TDM bus 130 and derives clock information therefrom in a conventional manner. The clock circuit 600 distributes this information in the form of clock signals at various rates including 2.048 MHz, 8 kHz and 50 Hz (corresponding to time intervals of 500 nsec, 125 µsec or 20 msec), all synchronized with each other, over a clock bus 615 to circuits 604-606, 608 and 611 to synchronize their operation with the TDM bus 130. The clock circuit 600 also distributes this information to the LAN bus interface 601 for bit time synchronization of the LAN bus 260. The operation of the TDM bus 130 is synchronized with the network 100 and the clock circuit 600 therefore synchronizes the functions of the various elements with the master clock of the network 100.

The adaptive synchronization circuit 611 uses the clock signals received from the clock circuit 600 to generate clock signals that are frequency synchronized with the 20 msec clock signals generated by the clock circuit 600 but are phase offset from them by amounts controlled by the processor 602. These offset clock signals are used to clock the functions of the processor 602. The generation and use of these offset clock signals is explained below. Physically, the circuits 611 and 600 can be implemented as a single device.

Circuit 611 also includes a current time counter 623. Counter 623 increments its count once per PCM sample tick, e.g., once every 125 µsec. This count is reset by each 50 Hz clock pulse from clock circuit 600, e.g., every 20 msec. Counter 623 therefore indicates the current time with respect to signals generated by clock circuit 600. A second portion of counter 623 maintains a modulo-8 count that is incremented by the 20 msec clock pulses that reset the 125 µsec count. Counter 623 provides its count to LAN bus interface 601 for use as a timestamp for receive frames 310.

The discussion now returns to processor 602 and its packet and frame processing functions (processing of the Layer 2 protocol is commonly referred to as frame processing, while processing of the Layer 3 protocol is commonly referred to as packet processing). The packets received by processor 602 from the LAN The functions performed on frames 310 received on bus 260 are illustrated in Figures 11-14. Processor 602 performs these functions every 20 msec. for each service line. Performance of different ones of these functions for a particular service line 612 is initiated by receipt of corresponding receive interrupt signals from LAN bus interface 601 and adaptive synchronization circuit 611.

As previously mentioned, when LAN bus interface 601 receives a frame addressed to the corresponding SPU 264, it appends a time stamp to the receive frame, stores the receive frame in buffer 620, and issues an interrupt to processor 602. When invoked by the receive interrupt signal from LAN bus interface 601 in step 900, processor 602 retrieves the receive frame from buffer 620 of LAN bus interface 601 in step 902. Processor 604 then performs ordinary level 2 processing, i.e., LAPD protocol, on the frame in step 904. This processing may include acknowledging receipt of the frame. Upon completion of level 2 processing, processor 604 checks control field 303 in step 906 to determine whether this is a level 2 only frame (e.g., a loopback frame). If so, then processing of the frame is completed and the processor 602 simply returns to its calling point in step 908. However, if this is not a level 2 only frame, i.e., its payload field 304 carries a level 3 protocol, then the processor 602 uses the frame's DLCI 302 to select from its memory, in step 910, the stored link state of the link to which the frame relates. Next, in step 911, the processor 602 checks the packet type field 321 of the received level 3 protocol to determine the packet type: traffic or signaling. If the field 321 identifies the packet as a signaling packet, this means that the packet carries cell-to-switch signaling information, i.e., signaling destined for the DCS 201. The Processor 602 therefore performs the indicated function in step 970. This may be one of three functions: updating the call state by either establishing or tearing down a call or adding or removing a second cell in soft handoff, inserting tones into the portion of the call intended for the telephone network, or performing initial clock synchronization (discussed in connection with Figure 17). Processor 602 then returns to its calling point in step 946. Voice/signaling packets 350 are sent and received at 20 msec intervals, while signaling-only packets 351 may be sent at any time according to the need to send signaling information.

If field 321 identifies the packet as a traffic packet, processor 602 performs timing adjustment and synchronization functions in step 912 to offset clock signals generated by circuit 611 from clock signals generated by circuit 600 by an amount determined by processor 602 or dictated by timing adjustment field 322 of the received packet. These are described in connection with Figure 18. Next, processor 602 checks the received Level 3 packet's voice/signaling type field 326 in step 914 to identify the type of information carried by the packet: voice only, voice plus signaling, or signaling only. If the traffic packet is a voice only packet, processor 602 checks the retrieved link state in step 916 to determine if the link is in the soft handoff state. If not, the processor 602 checks the frame's over-the-air CRC field 323 (which contains the result of a checksum calculated over the CDMA transmission between cell 202 and mobile phone 203) in step 918. If the over-the-air CRC is incorrect, this means that the packet carries erroneous information and the processor 602 therefore discards the packet in step 923 and returns in step 946. The vocoder 604 Loss of this traffic is masked. If the radio transmission CRC is correct in step 918, then the processor 602 checks the signal quality field 324 of the packet in step 919 to determine if the voice quality meets a predetermined threshold. If the voice quality meets the threshold, the processor 602 marks the packet as "good" by appending a command to it in step 920, stores the packet of voice information in the buffer 603 associated with the corresponding service line 612 in step 922, and then returns to its calling point in step 926. If the voice quality does not meet the minimum threshold, the processor 602 marks the packet as "bad" in step 921, stores the packet in the buffer 603 of the corresponding service line 612 in step 922, and then returns in step 946.

During the procedures just described, the processor 602 uses the contents of the sequence number field 320 of the received packet to detect and handle lost or misqueued packets in a conventional manner.

Returning to step 916, when the link is in the soft handoff state, processor 602 should receive two packets for the link every 20 msec, each from a different cell 202 but generally with the same information content. Processor 602 therefore checks in step 932 whether it has already received both duplicate packets. The duplicate packets are identified by having the same sequence number in field 320. If not, meaning that processor 602 has either received only one of the expected duplicate packets or has received packets from both cells but with different sequence numbers, processor 602 checks the sequence number of the packet just received in step 933 to determine whether its sequence number is greater than, equal to, or less than the expected sequence number. If the sequence number of the received packet is greater than the expected sequence number, processor 602 stores in Step 934 receives the received packet, updates the state of the associated link in step 935 to indicate that one of the expected packets has been received, and returns in step 946. Updating the link state in step 935 includes storing the contents of the power control field 325 of the receive packet. If the receive packet's sequence number is equal to the expected sequence number, processor 602 proceeds to step 918 et seq. to process the packet as previously described. If the receive packet's sequence number is less than the expected sequence number, processor 602 discards the received packet in step 936 and then returns in step 946. The loss of this traffic is again masked by vocoder 604.

Returning to step 932, if the processor 602 determines that it has received both expected packets, it updates the link state in step 938 to indicate this. This involves storing the contents of the power control field 325 of the received packet. It then retrieves the first expected packet received (now stored in a buffer 603) and compares the radio transmission CRC and signal quality indicators of both packets in step 940 to determine which packet is the better one. The processor 602 then checks the voice quality field of the better packet in step 941 to determine if the voice quality meets a predetermined threshold. If not, then the processor 602 marks the better packet as "good" by appending a command to it in step 943; if so, then processor 602 marks the better packet as "bad" in step 942. Processor 602 then discards the worse packet and stores the better packet in buffer 603 of the corresponding link channel in step 944. Processor 602 then returns in step 946.

Referring to Figure 12, after step 946, processor 602 checks whether processor 602 in step 950 is interrupted by a receive signal RX_INT_X is called for a particular (Xth) service line 612, the processor 602 retrieves the buffer 603 corresponding to that service line 612 in step 951 to determine if the buffer 603 is empty. If not, the processor 602 retrieves the contents of that buffer 603 and passes the retrieved contents to the vocoder 604 of that service line 612 in step 952. If the buffer 603 is empty, the processor 602 calls a function in the vocoder 604 of the corresponding service line 612 in step 953 to mask the loss of the speech segment carried by the discarded packet. The vocoder 604 masks the loss by producing at its output to line 605 PCM samples which it generates as a function of previously received packets. The processor 602 then returns to its calling point in step 954.

Returning to step 914, a traffic packet carrying signaling information is only encountered by processor 602 during a "soft handoff" since under normal circumstances signaling is sent directly to mobile telephone 203 from the cell 202 involved in a given connection. If the traffic packet carries only signaling information, then processor 602 proceeds to step 955 of Figure 13. There, processor 602 examines the remaining contents of voice/signaling type field 326 to determine the signaling direction forward and/or reverse. If the direction is forward, identifying the signaling as initiated by a cell 202 and destined for a mobile telephone 203, processor 602 simply stores the packet in step 956 and then returns in step 970. If both signaling directions are indicated, processor 602 stores the forward signaling in step 957 and then proceeds to step 958. If the direction is reverse, which identifies the signaling as initiated by a mobile telephone 203 and destined for cells 202, processor 602 checks in step 958 whether it has received signaling packets from both sides (ie, from both of the cells 202 involved in the "soft handoff". If not, then processor 602 stores the packet in step 960 and then updates the state of the corresponding link in step 962 to indicate that a signaling packet has been received from one side. Then processor 602 returns in step 970. If the check in step 958 indicates that signaling packets have been received from both sides, processor 602 updates the state of the corresponding link in step 964 to indicate this and then compares the radio transmission CRC and signal quality fields 323 and 324 of the two packets, respectively, in step 966 to determine which packet carries the better quality signals. Then processor 602 discards the poorer packet in step 968 and stores the better one and then returns in step 970.

Returning to step 914, if the processor 602 determines that the packet carries both voice and signaling information, it proceeds to step 985 of Figure 14 and performs signaling processing steps 985-998 of Figure 14, which are similar to steps 955-968 of Figure 13, and then proceeds to step 932 of Figure 11 to perform the voice processing steps.

Figure 15 shows the functions performed by processor 602 on traffic frames (segments of speech information) received from vocoders 604. Processor 602 performs these functions every 20 msec for each service line 612. Performance of the functions for a particular service line 612 is also interrupt-controlled by receipt of a corresponding transmission interrupt signal provided by adaptive synchronization circuit 611.

When called by a transmission interrupt signal TX_INT_X to start processing for a specific (X-th) service line 612 in step 1200, the processor 602 checks the stored connection state of the service line 612 served by this service line 612. connection in step 1202 to determine if the connection is in soft handoff. If not, processor 602 accesses vocoder 604 of served service line 612 and requests therefrom in step 1227 a traffic frame of full bit rate encoded connection information. Upon receiving a traffic frame from this vocoder 604 in step 1228, processor 602 formats the traffic frame in the Layer 3 protocol in step 1230. This includes prepending the connection traffic with a sequence number and a traffic type. Thereafter, processor 602 encapsulates the formatted traffic frame in the LAPD frame format in a conventional manner in step 1232 to form LAPD frame 300 (see Figure 7). This includes retrieving the DLCI associated with the mobile's designated call direction, which identifies a particular channel element 245 of a particular cell 202 (see Figure 3) serving the call, and including it in the LAPD frame 300. The processor 602 then uses this DLCI to find the card and port addresses 311 and 312 corresponding to that DLCI in a table, and prepends the found addresses 311 and 312 to the LAPD frame 300 to form a modified LAPD frame 310 in step 1234 (see Figure 8). The processor 602 passes the frame 310 to the LAN bus interface 601 in step 1236 for transmission on the LAN bus 260. The processor 602 then returns to its calling point in step 1238.

Returning to step 1202, if the processor 602 determines that the connection is in the soft handoff state, in step 1204, it checks the stored connection state of the connection to determine if any forward signaling is stored for that line. Forward signaling would have been received only by the cell 202 (referred to as routing cell 202) that handled the connection and stored in step 956 or 957 of Figure 13 or step 986 or 987 of Figure 14. If no forward signaling is stored, the processor accesses 602 accesses the vocoder 604 of the service line 612 being served and requests therefrom a traffic frame with full-rate encoded communication information in step 1206. However, if forward signaling is stored, the processor 602 must reserve space in a packet for the forward signaling information and therefore accesses the vocoder 604 and requests therefrom a traffic frame with only partial-rate encoded communication information in step 1208.

The vocoder 604 typically provides traffic frames with full rate encoded information and may not be able to immediately respond to a request for a traffic frame with partial rate encoded information. Furthermore, during a pause in the call, a partial rate encoded traffic frame may be provided even if a full rate encoded traffic frame has been requested. The processor 602 checks for this condition in step 1218.

Having received a traffic frame from vocoder 604 in step 1209, processor 602 copies the traffic frame in step 1210 so that it can send copies to the two cells 202 participating in the soft handoff. Then, in step 1212, processor 602 retrieves power control information from the two cells 202 participating in the soft handoff, which will be stored in steps 935 and 938 of Figure 11, swaps them so that each of the two cells 202 is sent the power control information received from the other of the two cells 202, and inserts the swapped information into the duplicated packets as power control field 325 in step 1212. Next, in step 1214, the processor 602 checks the state of the connection to determine whether reverse signaling for the connection was received and stored in step 968 of Figure 13 or step 998 of Figure 14. If reverse signaling is available, the processor 602 appends it to both duplicated packets in step 1216. After step 1216 or if no reverse signaling is available, processor 602 checks in step 1218 whether it has been provided with a frame of full rate encoded or partial rate encoded information by vocoder 604. If the traffic frame is full rate encoded, it has no room for forward signaling information and processor 602 therefore proceeds to steps 1230 et seq. to format, packetize and transmit both duplicated packets. Packetization in step 1234 means including a different DLCI in the frame protocol 300 of each duplicated packet so that the two packets each go to a different cell 202 participating in the soft handoff. Returning to step 1218, if the traffic frame is sub-rate encoded, processor 602 checks the state of the link in step 1220 to determine if forward signaling for the link was received and stored in step 956 of Figure 13 or in step 986 of Figure 14. If forward signaling is available, processor 602 appends it to both duplicated packets in step 1222. After step 1222, or if forward signaling is not available, processor 602 proceeds to steps 1230 et seq.

The synchronization of the operations of cell 202 and SPU 264 will now be explained in more detail in connection with Figures 16-22.

Figure 19 illustrates the scenario for the initial timing settings for traffic flow from the network 100 to mobile phones 203. As previously mentioned, the operations of all mobile phones 203 and all channel elements 245 of all cells 202 are controlled and synchronized with a common timing signal, which may be a signal broadcast by a navigation satellite. Each cell 202 derives therefrom a 20 msec cell timing signal 1000 which controls each channel element 245 involved in a connection to make a transmission to the corresponding mobile phone 203 at time 1300 every msec. For a given connection, a programmed constant offset (ie, a offset between the rising edge of the cell clock 1000 and the time Tx 1300), which can be zero. This constant offset influences the relative positions of the signals 1304, 1307, 1308 and 1309 by the amount of the said offset.

In order to transmit call traffic at time 1300, a channel element 245 must receive that call traffic at least some minimum amount of time before time 1300 at a time tmin 1301. Channel element 245 preferably receives the information for transmission within a time window 1302 that is slightly after time 1300 of the previous transmission and slightly before time 1301 of the current transmission. Window 1302 therefore provides some margin for minor timing variations. However, when a call is established, it is uncertain when the channel element 245 handling the call will receive a packet of call traffic for transmission from SPU 264. The reason for this, as previously mentioned, is that the operations of the mobile telephone exchanges 201 are controlled by a clock other than that of the cells 202, which is not synchronized with cell clock 1000 but is independent of it. Furthermore, the reception time is also made uncertain by other factors such as differences in distances between mobile telephone exchanges 201 and different cells 202 and the transmission of different traffic values between them - and thereby causing different propagation times between them. Therefore, when a connection path is first established between a channel element 245 and an SPU 264 and zero traffic begins to flow between them, the channel element 245 may receive packets from the SPU 264 at times 1303 that are outside the windows 1302 and, in the worst case, after the times tmin 1301. If this is the case, the corresponding channel controller 244 of the channel element sends the SPU 264 a signaling packet indicating the need to adjust the transmission time of packets from the SPU 264 and also indicating the amount of time by which this transmission time may be adjusted. in order to ensure that the reception time of the packets at the channel element 245 is placed within the windows 1302.

Figure 16 illustrates the timing adjustment functions performed at cell 202. They represent a routine executed by the processor when a packet is received at multi-controller 244. When called at step 1001, the routine checks at step 1002 whether the received packet is the first traffic packet received for the connection. If so, the routine compares the time of receipt of the packet at step 1004 to a window 1302 (the definition of which is stored in multi-controller 244) to determine at step 1006 when the packet was received relative to window 1302. If the packet was received substantially in the middle of window 1302, no timing adjustment is necessary and the routine simply returns to its point of invocation at step 1022. If the packet was received too early, the routine causes a cell switch signaling packet to be transmitted to the processor 602 of the SPU 264 handling the connection in step 1008 requesting that the processor 602 delay the TX_INT_X interrupt time for that connection by a time also specified in the packet which will move the reception time substantially to the middle of the window 1302. Conversely, if the packet was received too late, the routine causes a cell switch signaling packet to be transmitted to the processor 602 in step 1010 requesting that the TX_INT_X interrupt time for that connection be advanced by a specified time. The routine then returns to its calling point in step 1022.

Alternatively, the routine need not simply respond to the first traffic packet received, but may calculate an average time of required timing adjustment based on the receipt of a plurality of received traffic packets.

Packet reception times 1303 at channel element 245 correspond to packet transmission times 1304 at SPU 264. As As previously mentioned, transmission of packets to channel element 245 from SPU 264 is initiated by transmit interrupt signals TX_INT_X issued by adaptive synchronization circuit 611 to processor 602. Adjusting the packet reception times at channel element 245 by a certain amount consequently requires adjusting the TX_INT_X signals at circuit 611 by the same amount. Thus, when processor 602 receives the above-mentioned signaling packet from channel element 245, it responds in step 970 of Figure 11 by instructing adaptive synchronization circuit 611 to adjust the TX_INT signal for the corresponding service line 612 by the specified amount. Circuit 611 complies and shifts the transmit interrupt signal by the specified amount of time designated as 1310 in Figure 19. The packet transmission time is thus shifted from times 1304 to times 1305 at the SPU 264, which corresponds to the packet reception times 1306 at the channel element 245. The packet reception times 1306 are within the windows 1302.

However, in order to transmit a packet at a given time, processor 602 must receive the traffic frame (portion of call traffic) contained in that packet from vocoder 604 at a time before the transmission time. Packet transmission times 1304 correspond to frame reception times 1307, which in turn correspond to traffic frame transmission times 1308 of vocoder 604, while shifted packet transmission times 1305 correspond to shifted traffic frame reception times 1311, which in turn correspond to traffic frame transmission times 1309 of vocoder 604. Processor 602 must therefore cause vocoder 604 to shift its traffic frame transmission times from times 1308 to times 1309.

The vocoder 604 uses the output of an internal output clock 622 to clock its traffic frame transmissions. The clock 622 of an X-th service line 612 is initially synchronized with clock input signals received from the clock circuit 600. The processor 602 sends a command to vocoder 604 to adjust the offset of its output clock signals 622 from the clock input signals of circuit 600 by the aforementioned time period 1310 specified in the signaling packet received by processor 602 from channel element 245. This is done by vocoder 604, which thereby shifts its traffic frame transmission times from times 1308 to times 1309. The net result is that the asynchronous operations of channel element 245 and service line 612 and processor 602 have become synchronized with each other.

Figure 17 diagrammatically illustrates the response scenario of processor 602 to receiving the clock adjustment signaling packet from cell 202. Upon determining in step 1050 that the received signaling packet requests that clock adjustments be made, processor 602 examines the packet contents in step 1052 to determine the direction in which the clock signals are to be shifted. If they are to be delayed, processor 602 sends a command to adaptive synchronization circuit 611 in step 1054 to delay subsequent interrupt signals TX_INT_X by the amount of time specified in the packet. Also, in step 1056, processor 602 sends a command to vocoder 604 to increase the offset of its output clock 622 from signals from clock 600 by the same amount of specified time and then returns in step 1062. If the clock signals are to be advanced in time, processor 602 sends a command to adaptive synchronization circuit 611 in step 1058 to advance subsequent interrupt signals TX_INT_X by the amount of time specified in the received signaling packet. Also, in step 1060, processor 602 sends a command to vocoder 604 to decrease the offset of its output clock 622 from signals from clock 600 by the same amount of specified time and then returns in step 1062.

Figure 20 presents the scenario for initial timing settings for the traffic flow of mobile radiotelephones 203 to network 100. As previously mentioned, mobile telephones 203 and cells 202 are synchronized with one another. A clock corresponding to cell clock 1000 (derived by mobile telephone 203 from traffic received by cell 202) causes mobile radiotelephone 203 to make a transmission every 20 msec to channel element 245 handling the call, whereby channel element 245 receives these transmissions at times 1400 and transmits them in packets to SPU 264 at times 1403. Packet send times 1403 on channel element 245 correspond to packet receive times 1404 on processor 602 of SPU 264. Receive times 1400 are offset relative to send times 1300 by the amount of a programmed constant offset on cell 202 relative to cell clock 1000. Thus, an offset in send times 1300 results in an equal offset in receive times 1400. This offset is compensated by the mechanisms described here.

Reception of packets from channel element 245 for a particular (Xth) service channel 612 is triggered at processor 602 by a receive interrupt signal RX_INT_X for that service channel 612 generated by adaptive synchronization circuit 611. Reception of the packets must precede transmission of the paging traffic frames contained in the packets to vocoder 604 by some minimum time to allow processor 602 sufficient time to process the packets. Vocoder 604 initially expects to receive traffic frames at times 1408 corresponding to traffic frame transmission times 1406 from processor 602. Therefore, in order to transmit traffic frames to vocoder 604 at times 1406, processor 602 must receive corresponding packets from channel element 245 no later than times tmjfl 1401. Processor 602 preferably receives each packet within a time window 1402 that exists slightly after the send time 1406 of the previous frame transmission to vocoder 604 and slightly before the time tmin 1401 of the current frame transmission. Thus, window 1402 provides some margin for minor timing variations.

However, when a connection is established, it is uncertain when processor 602 will receive a payload packet from channel element 245, for the same reasons that it is uncertain when channel element 245 will receive a packet from processor 602, as discussed above. Therefore, when a connection path is first established between a channel element 245 and an SPU 264 and zero traffic begins to flow between them, packets from channel element 245 may be received by processor 602 at times 1404 that are outside windows 1402 and, in the worst case, after times tmin 1401. Processor 602 cannot change the times 1403 at which channel element 245 transmits packets and therefore cannot change the times 1404 at which it receives those packets; processor 602 can only change times 1406 when it transmits frames to vocoder 604. Therefore, if times 1404 are outside of windows 1402, processor 602 determines a time period 1410 by which it must shift its frame transmission time to vocoder 604 to ensure its packet reception times 1404 fall within windows 1402. Processor 602 then instructs adaptive synchronization circuit 611 to shift the receive interrupt signal RX_INT_X for the corresponding service line 612 by the specified amount. Circuit 611 obeys and shifts this receive interrupt signal by the specified time period 1410. Frame transmission times from processor 602 to vocoder 604 are therefore shifted from times 1406 to times 1407, shifting packet reception times 1404 at processor 602 into windows 1402. However, in order to shift its frame transmission times from times 1406 to times 1407, the processor 602 must cause the vocoder 604 to shift its frame reception times from times 1408 to times 1409. The vocoder 604 uses the output of an internal input clock 621 to clock its frame receptions. Like the output clock 622, the input clock 621 is synchronized with the clock inputs 600. The processor 602 therefore sends a command to the vocoder 604 to offset its input clock signals 621 from the clock input signals 600 by the time period 1410 mentioned above. This is accomplished by the vocoder 604 which thereby shifts its frame reception times from times 1408 to times 1409. The net result is again that the asynchronous operations of the channel element 245 and the service line 612 and the processor 602 have become synchronized with each other.

The timing adjustment functions just described are performed by processor 602 in step 912 of Figure 11 and illustrated in Figure 18. Upon beginning to perform the timing adjustment function in step 1070, processor 602 determines from the retrieved link state and the type of packet received in step 1072 whether the received packet is the first traffic packet for the link. If so, processor 602 compares the packet's receive timestamp (attached to the packet by LAN interface 601) to a window 1402 (the definition of which is calculated and stored by processor 602 for each link it handles) in step 1073 to determine when the packet was received relative to window 1402 in step 1074. If the packet was received substantially in the middle of window 1302, no timing adjustment is necessary and processor 602 proceeds to step 1090. If the packet was received too early, processor 602 instructs adaptive synchronization circuit 611 in step 1075 to advance subsequent interrupt signals RX_INT_X by the amount of time determined by processor 602 as necessary to move the reception time substantially to the center of window 1402. Also, processor 602 in step 1076 sends a command to vocoder 604 to increase the offset of its input clock 621 from clock signals 600 by the same amount of specified time. Conversely, if the packet was received too late, processor 602 instructs adaptive synchronization circuit 611 in step 1077 to advance subsequent interrupt signals RX_INT_X by the amount of time determined by processor 602 as necessary to move the reception time substantially to the center of window 1402. Processor 602 determines it is necessary to move the receive time substantially to the center of window 1402. Also, in step 1078, processor 602 sends a command to vocoder 604 to decrease the offset of its input clock 621 from clock signals 600 by the same amount of specified time. After step 1076 or 1078, processor 602 proceeds to step 1090 (described below).

As the call progresses, changes in system traffic load or drift between the master clock to which cells 202 are synchronized and the master clock to which mobile telephone switches 201 are synchronized may cause packet reception times 1306 at channel elements 245 to drift out of windows 1302, as exemplified in Figure 21, and packet reception times 1404 at processor 602 of SPU 264 to drift out of windows 1402, as exemplified in Figure 22. The drift due to changes in system traffic load will generally be in the same direction with respect to times 1306 and 1404; a wander that advances time 1306 with respect to window 1302 (shown in Figure 21) will typically also advance time 1404 with respect to window 1402 (not shown), while a wander that resets time 1404 with respect to window 1402 (shown in Figure 22) will typically also reset time 1306 with respect to window 1302 (not shown). Conversely, wandering due to asynchrony between the master clocks will generally occur in opposite directions.

The departure of the times 1306 from the windows 1302 is detected by the corresponding multiple control unit 244 of the channel element. Its reaction to this is shown in Figure 16. When a packet is received at the multiple control unit 244, the routine of Figure 16 is called in step 1001 and checked by this in step 1002 whether the received packet is the first for the connection. Since the connection is in the talk state, this will not be the first traffic packet received and the routine continues at step 1014. There, the routine compares the time of receipt of the packet to window 1302, the same as in step 1004, to determine when the packet was received relative to window 1302 in step 1016. If the packet was received within window 1302, no timing adjustment is necessary and the routine simply returns at step 1022. If the packet was received before window 1302 occurred, the routine causes the next traffic packet for this connection sent to processor 602 of the SPU 264 handling the connection to carry in its timing adjustment field 322 a request to delay the time of interrupts TX_INT X for this connection by one tick (e.g., one PCM voice sample time). Conversely, if the packet was received after the occurrence of window 1302, the routine causes the next traffic packet for this connection to carry in its timing field 322 a request to processor 602 to advance the time of interrupts TX_INT_X for this connection by one tick in step 1020. After step 1018 or 1020, the routine returns to its calling point in step 1022.

Upon receipt of the traffic packet, the processor 602 proceeds to perform the corresponding adjustment in step 912 of Figure 11. Migration of the times 1404 from the windows 1402 is detected by the processor 602 itself. The processor 602 notices the need for adjustment and the direction of adjustment and also proceeds to perform the corresponding adjustment tick by tick in step 912 of Figure 11.

If a change in the timing of the activity of the processor 602 advances the packet sending times 1305 from times 1305 to times 1505 and therefore advances the packet receiving times 1306 with respect to window 1302, new packet receiving times 1506 result, which are as shown in Figure 21 shown, again lie within the windows 1302. If, by changing the timing of the activity of the processor 602, the windows 1402 and frame transmission times 1406 are advanced with respect to times 1404, new frame transmission times 1606 and packet reception times 1404 result, which again lie within the windows 1402 as shown in Figure 22.

The shifting of the TX_INT_X and RX_INT_X signals output by circuit 611 requires a corresponding shifting of the signal outputs of clocks 621 and 622 of vocoder 604, thereby changing the traffic frame transmission times of vocoder 604 from times 1309 to times 1509 and the traffic frame reception times of vocoder 604 from times 1409 to times 1609 in the examples of Figures 21 and 22, thereby resynchronizing the operations of vocoder 604 with the time-shifted operations of processor 602. However, at the moment of resynchronization, vocoder 604 must present a frame of link traffic to processor 602 after vocoder 604 has had time to collect either 159 or 161 PCM samples from circuit 605 corresponding to a period of 20 msec instead of the normal 160, and output a frame of link traffic to circuit 605 within a period of either 159 or 161 PCM samples instead of the normal 160, depending on whether the setting to advance or delay the interrupt signals occurs. To compensate for this condition, at the same time as it instructs the circuit 611 to effect the shifts of its TX_INT_X and RX_INT_X signals for this service line 612, shown in Figures 21 and 22, respectively, the processor 602 instructs the vocoder 604 of this same service line 612 to drop a PCM sample byte from its PCM output signal and to generate an additional PCM sample byte at its PCM input. This is what the vocoder 604 does, with the result that the traffic frame input and output activities of the vocoder 604 again become synchronous with the PCM sample output and input activities, respectively.

When wandering in the opposite direction to that shown in Figures 21 and 22, the steps taken to compensate for wandering are the reverse of those described for Figures 21 and 22. Specifically, processor 602 instructs circuit 601 to delay its interrupt signal outputs TX_INT_X and RX_INT_X for that service line 612 by one PCM sample period and instructs vocoder 604 to generate an additional PCM sample byte at its PCM output and to drop a PCM sample byte from its PCM input.

These activities of processor 602 are diagrammatically illustrated in steps 1080 et seq. of Figure 18. As previously mentioned, when processor 602 begins the timing activities of step 912 of Figure 11 in step 1070, processor 602 checks to see if the packet just received is the first traffic packet on the connection. While the connection is established, a receive packet will not be the first packet received and processor 602 therefore proceeds to step 1080. There, processor 602 again compares the timestamp of the receive packet to receive window 1404 to determine when the packet was received relative to the window in step 1081. If the packet was received within window 1404, no timing adjustment is necessary and processor 602 proceeds to step 1090. If the packet was received before window 1404, processor 602 instructs adaptive synchronization circuit 611 in step 1082 to advance the RX_INT_X signal for the corresponding service line 612 by one tick and instructs vocoder 604 in step 1083 to decrease the offset of its input clock 621 by one tick. This is accomplished by vocoder 604 by causing clock 621 to reset after a count of 159 instead of the usual count of 160. Nevertheless, vocoder 604 still receives a complete traffic frame of incoming link traffic with the equivalent of 160 PCM sample bytes of information. The vocoder 604 therefore discards one of these sample bytes to mask the time resynchronization at its PCM output.

Returning to step 1081, if the packet is found to have been received after window 1404, processor 602 instructs adaptive synchronization circuit 611 in step 1084 to delay the RX_INT_X signal for the corresponding service line 612 by one tick and instructs vocoder 604 in step 1085 to increase the offset of its input clock 621 by one tick. Vocoder 604 accomplishes this by causing clock 621 to reset after a count of 161 instead of the usual count of 160. Nevertheless, vocoder 604 still receives a traffic frame of incoming traffic with the equivalent of 160 PCM sample bytes of information. The vocoder 604 therefore generates an additional sample byte to mask the time resynchronization at its PCM output.

After steps 1083 and 1085, respectively, the processor 602 proceeds to step 1090. There, the processor 602 examines the clock adjustment field 322 of the received traffic frame to determine what, if any, clock adjustment has been requested by the cell 202 handling the connection. If an adjustment has been requested, the processor 602 instructs the adaptive synchronization circuit 611 to adjust the time of occurrence of the interrupts TX_INT_X for the corresponding service line 612 of the connection by one tick in the requested direction, in step 1091, and instructs the vocoder 604 to adjust the offset of its output clock 621 by one tick in the same direction, in step 1092. The vocoder 604 does this by causing the clock 621 to reset after a count of 159 or 161 instead of the usual count of 160. As a result, the vocoder 604 collects either 159 or 161 PCM bytes of outgoing traffic samples for input to processor 602 in a frame comprising 160 PCM sample bytes. To mask the temporal resynchronization at its output to processor 602, vocoder 602 first generates an additional PCM sample and second discards one of the PCM samples. After step 1092, the timing adjustment activities are complete and processor 602 returns to the connection processing activities of Figure 11 in step 1093. Alternatively, timing adjustments may be made in multiples of one 125 µsec ticks to achieve higher speed synchronization. Also, a combination of multiple tick and one tick adjustments (in different 20 msec cycles) could be used to control the speed at which synchronization can be achieved. Furthermore, trial adjustments (ie with several 125 µsec ticks) can be performed to make significant synchronization changes during a connection. The said large adjustments are advantageously performed during periods of low voice activity.

At the beginning of a soft handoff, a channel element 245 of a second cell 202 begins to service the connection in parallel with the channel element 245 of a cell 202 that has been handling the connection alone up to this point. It is not known in advance whether the packet reception times 1306 at the second channel element 245 will fall within or outside the windows 1302 (see Figure 19) or whether packet reception times 1404 of packets sent by the second channel element 245 will fall within or outside the windows 1402 (see Figure 20) at the processor 602, just as they did when the connection was first established. However, if the reception times 1306 and 1404 fall outside the windows 1302 and 1402, respectively, for the second channel element 245, the timing adjustment method of Figures 19 and 20 that was used when the connection was first established cannot now be used. This is because the connection is now an established and ongoing connection, and the application of this procedure would cause a noticeable disturbance - an audible "glitch" - in the conversation. Consequently, the more gradual but effectively "glitch-free" timing adjustment procedure of Figures 21 and 22 is used to attempt to shift the receive times 1306 and 1404 into the windows 1302 and 1402, respectively, for the second channel element 245. Several iterations of this adjustment may be necessary to achieve the desired effect.

It is important to note, however, that the adjustment of Figures 21 and 22 affect the receive times 1306 and 1404 for both channel elements 245 handling the connection. As a result, it is possible that an adjustment attempting to move times 1306 and 1404 into windows 1302 and 1402 for the second channel element 245 may push times 1306 and 1404 out of windows 1302 and 1402 for the first channel element 245.

It is most important that the times 1306 and 1404 of the two channel elements 245 do not lag behind (i.e., occur after) their respective windows 1302 and 1402. Conversely, times 1306 and 1404 that lead (i.e., occur before) their respective windows 1302 and 1402 can be compensated for by buffering the prematurely received packets at channel element 245 and SPU 264. Therefore, during soft handoff, if one channel element 245 reports a leading time 1306 while the other channel element 245 reports a lagging time 1306, the clock adjustment requests of the channel element 245 reporting the leading times 1306 are ignored and the processor 602 responds only to the requests of the other channel element 245 reporting lagging times 1306.

It is conceivable that differences in the runtimes between the processor 602 and the two channel elements 245 involved in the soft switching are so are large, that packets sent by both channel elements 245 during the same clock cycle of the cell clock 1000 are received at the processor 602 during different clock cycles of the receive interrupt clock RX_INT_X of the processor 602 for that channel element 612, and that duplicate packets sent by the processor 602 during the same clock cycle of the transmit interrupt clock TX_INT_X to both channel elements 245 participating in the soft handoff are received by those channel elements 245 during different clock cycles of the cell clock 1000. Associating the received packets with the correct clock cycles is the purpose of the sequence numbers carried by the sequence number field 320 of the traffic frames 350 (see Figure 9). The assignment is performed in steps 932-936 of Figure 11.

As already indicated, sequence numbers used by channel elements 245 are calculated from clock cycles of cell clock 1000 and therefore have a defined relationship with them. Therefore, on any clock cycle of cell clock 1000, all channel elements 245 transmit packets with the same sequence number. As a result, by comparing the sequence numbers of two received packets, processor 602 can immediately determine whether both packets correspond to the same clock cycle of clock 1000 and, if not, what their relative sequence is.

In the opposite direction of packet flow from processor 602 to channel elements 245, there is no defined relationship between the sequence number and the clock cycle of cell clock 1000. At the beginning of the soft handoff, the channel element 245 that has been handling the connection thus far causes a message (HANDOFF_REQ; see discussion of Figure 27 below) to be sent to the channel element 245 that is now beginning to handle the connection, in which message the number of a most recent clock cycle of cell clock 1000 and the sequence number of a packet that the first channel element 245 received during that clock cycle are reported. Since sequence numbers are sequential, the second channel element 245 can determine from this received information It is easy to calculate which sequence numbers are associated with which subsequent clock cycles of the cell clock 1000. The second channel element 245 thus determines the clock cycle of the cell clock 1000 corresponding to a received packet.

Referring now to Figures 23-35, it is explained how connections are set up, forwarded, and torn down in the system of Figure 2. The activities illustrated occur as a result of the exchange of packetized Level 3 signaling messages, for example, between pairs of elements, e.g., SPU 264 to cells 202, cell 202 to ECP complex 134, or ECP complex 134 to DCS controller 261. The figures represent timing relationships for message exchanges only between pairs of elements and not across pairs of elements. It is assumed that all messages to and from ECP complex 134 flow over control links 108; it is assumed that all packets between channel elements 245 and service lines 612 are frame forwarded over lines 207 and 210.

Figure 23 shows the control signaling for establishing a packet-switched connection path for a call originating from a mobile telephone 203. The mobile telephone 203 initiates the call by transmitting an ORIGINATION signal (e.g., one or more digital messages) conveying the called telephone number on an access channel. Wireless transmissions or wireless reception of signals are indicated in the figures by a vertical segment of a signal arrow. The ORIGINATION signal is received by the channel element 245 designated as the CDMA access channel in one of the cells 202, which passes it in a message to its multiple control unit 244, which forwards it to the controller 241 of its cell 202. Each controller 241 allocates a free CDMA radio channel to carry the connection and then passes the message, together with the identification of the corresponding channel elements 245 of the assigned channel, to the ECP complex 134 in the usual manner.

The ECP complex 134 receives the CELL_ORIGINATION message and selects a DCS 201, a CIM 209, an SCM 220, and a service line 612 and a group of lines 106 of the selected speech encoder module 220 to process the call. The ECP complex 134 then sends an MSC_FS_ASSIGNMENT message to the controller 241 of the call originating cell 202 with a DLCI of the selected service line 612. The ECP complex 134 also sends a SETUP message with the called telephone number and identification of the selected module 220, groups of lines 106, and service line 612 to the DCS controller 261 controlling the selected module 220.

The controller 241 receiving the MSC_FS_ASSIGNMENT message forwards the message to the multi-controller 244 of the selected channel element 245. The multi-controller 244 transmits the information contained in the message to the channel element 245 selected for connection processing. The selected channel element 245 configures itself to process the connection and then sends a FS_CONNECT packet 351 to the selected service line 612 using the frame forwarding method to transport the packet over the channels of the interconnect devices. Packet 351 uses the received DLCI of the selected service line 612 as the packet address in field 302 and transmits the DLCI of the selected channel element 245 in its data field 304.

When the processor 602 serving the selected service line 612 receives the FS_CONNECT packet, it sends an FS_ACK packet 351 back to the selected channel element 245 as an acknowledgement of receipt of the FS_CONNECT packet, using the DLCI contained in field 304 of the FS_CONNECT packet as the packet address in field 302 of the FS_ACK packet. For example, at this time the processor 602 also sends all DLCI corresponding to the selected service line 612 to the cell 202. The processor 602 performs these tasks as part of the LAPD processing in step 904 of Figure 11. Thereafter, the processor 602 stores the transmitted DLCI of the selected channel element 245 as part of the connection state associated with the selected service line 612 and marks the connection state as corresponding to an active connection. A connection now exists between the selected channel element 245 and the service line 612. Next, the multiplex controller 244 of the selected channel elements 245 responds with a FS_CLOCK_ADJUST packet communicating the initial clock adjustment information to the processor 602 serving the selected service line. This packet was discussed in connection with Figure 16, steps 1001-1010. The processor 602 responds by returning a FS_ACK packet to the multiplex controller 244 and processing the receive packet in the manner discussed in connection with Figure 17. A connection path now exists between channel element 245 and service line 612 and they begin exchanging null traffic packets every 20 msec until connection traffic becomes available. The selected channel element 245 responds to receipt of the second FS_ACK packet by causing a CHANNEL_CONFIRMATION message to be sent from its cell's controller 241 to ECP complex 134 to inform it of the establishment of this end of the connection.

The DCS controller 261 receiving the SETUP message responds by causing the controller 231 of the selected SCM 220 to seize a line 106 (DS0 channel) of the designated groups of lines 106 and to output the pulses of the called telephone number on the seized line 106. The selected line 106 corresponds to a particular time slot on the TDM bus 130. The controller 261 also causes the translation and maintenance processor 609 of the speech processing unit 264 containing the selected service line 612 to connect the above-mentioned DS0 channel from the TDM bus 130 via the TDM bus interface 608 to the time slot of the collective bus 607 assigned to the selected service line 612, thereby instructing the processing of the relevant connection to this service line 612. The controller 261 then sends a CONNACK message to the ECP complex 134 to inform it of the successful establishment of this end of the connection. When response monitoring is received by the controller 231 from the telecommunications equipment of the network 100 over the selected line 106, it notifies the DCS controller 261, which in turn sends an ANSWER message to the ECP complex 134 to inform it of the connection establishment. The connection is now complete through the system of Figure 2 and connection traffic can flow between the selected channel elements 245 over the service line 612 and line 106 to and from the telecommunications equipment of the network 100 and the connection destination.

Figure 24 shows the control signaling for establishing a connection path for a call originating from the public telephone network 100. The network 100 initiates the call by seizing a line 106 and then outputting the pulses of the digits of the called telephone number in the usual manner. The controller 231 of a speech encoder module 220 serving this line 106 detects the seizure in the appropriate time slot of the line on the TDM bus 130 and again collects the dialed digits in the usual manner and then notifies the DCS controller 261. The controller 261 in turn notifies the ECP complex 134 by sending it an INCALL message. The INCALL message conveys the called telephone number and the identifiers of the module 220 and the line 106.

The ECP complex 134 responds to the INCALL message by broadcasting an MSC_PAGE_REQUEST message to all cells 202 in the system of Figure 2. The MSC_PAGE_REQUEST message identifies the called mobile telephone 203 (e.g., conveys the called telephone number).

The controller 142 of each cell 202 responds to the message MSC_PAGE_REQUEST by sending the message MSC_PAGE_REQUEST is transmitted to a CDMA access channel element 245 via the multiplex controller 244. The access channel element 245 responds by paged the called mobile telephone 203 in the manner specified for the CDMA arrangement.

When the called mobile telephone 203 responds by transmitting a RESPONSE signal, the signal is received by one or more of the calling channel elements 245 and each forwarded to its corresponding multiple control unit 244. The multiple control units 244 forward the messages to controllers 241 of their corresponding cells 202. The controllers 241 of all cells 202 continuously exchange messages (not shown) to update each other on the respective state of their databases for existing and pending connections. The controllers 241 of the corresponding cells 202 determine from these messages which cell 202 is best suited to handle the connection. The controller 241 of the selected cell 202 then sends a CELL_PAGE_RESPONSE message to the ECP complex 134 to inform the complex 134 of the selection of that cell for processing the connection.

The ECP complex 134 receives the CELL_PAGE_RESPONSE message and selects a service line 612 of the module 220 to which the call will be connected to process the call at the other end of the call path. The ECP complex 134 then sends an MSC_FS_ASSIGNMENT message to the controller 241 of the selected cell 202. The message is the same as that described for the mobile call origination and elicits the same response - namely, a packet exchange sequence FS_CONNECT, FS_ACK, FS-CLOCK_ADJUST and FS_ACK between cell 202 and SPU 264 followed by a CHANNEL_CONFIRMATION message from the cell 202 to the ECP complex 134 as described for Figure 23. Also, the ECP complex 134 sends a TONE_REQ message to the DCS controller 261, which controls the module 220 to which the link is connected. The controller 261 responds by causing the Controller 231 of module 220 applies ringing to line 106 which connects to and from telecommunications equipment of network 100.

After sending the CHANNEL_CONFIRMATION message to the ECP complex 134, the selected channel element 245 transmits RINGING to the called mobile telephone 203. If the called mobile telephone 203 responds with an ANSWER message, the selected channel element 245 causes an ANSWER message to be transmitted from the controller 241 of its cell to the ECP complex 134. The ECP complex 134 responds by sending an ACCEPT message to the DCS controller 261 of the module 220 to which the call is connected. The message conveys the identification of the service line 612 that has been selected to handle the call. Controller 261 responds by causing controller 231 to remove the ringing tones from the connection and then causing a connection to be established between the DS0 channel carrying the connection on TDM bus 130 and the selected service line 612 in the manner described for a mobile originating connection. Controller 261 then sends a CONNACK message to ECP complex 134 to inform it of the successful establishment of this end of the connection. The connection path is now completely established through the system of Figure 2 and packets carrying call traffic can flow over service line 612 between the selected channel element 245 and the source of the call.

Figure 25 shows the control signaling for disconnection of the connection initiated by the mobile telephone 203. The mobile telephone 203 initiates the disconnection of an existing connection in which it is participating by transmitting a HANGUP signal. This signal is received by the channel element 245 handling the connection. The channel element 245 responds by sending an FS_REMOVE packet 351 to the service line 612 handling the connection to inform it of the disconnection of the connection.

Processor 602 responds to the FS_REMOVE packet by returning to channel element 245 an FS_ACK packet 351 as part of the protocol processing of the FS_REMOVE packet and updating the link state for the service line 612 handling the link to indicate that the link has been shut down. Traffic for the link is now suspended between channel element 245 and service line 612 and channel element 245 causes its cell controller 241 to send a RELEASE_MSC message to ECP complex 134 to inform it of the shutdown of that end of the link path.

The ECP complex 134 responds by sending a CLEAR message to the DCS controller 261 of the speech encoder module 220 handling the connection and by sending an MSC_RELEASE_ACK message to the controller 241 of the cell 202 that previously handled the connection to inform it that the channel element 245 that previously handled the connection is now free and available to handle a new connection. The controller 261 responds to the CLEAR message by causing the controller 231 of the module 220 to release the call-carrying line 106 and the re-evaluation and maintenance processor 609 of the voice processing unit 264, which contains the service line 612 handling the call, to shut down the DS0 channel carrying the call from the time slot of the bus 607 assigned to that service line 612. The controller 261 then sends a CLEAR_ACK message to the ECP complex 105 to inform it that this end of the call path has also been shut down.

Figure 26 shows control signaling for call disconnection initiated by the public switched telephone network 100. The network 100 releases the line 106 carrying the call. The release is detected by the controller 231 of the speech encoder module 220 handling the call, which notifies the DCS controller 261, and the controller 261 in turn notifies the ECP complex 134 by sending him a DISCONNECT message.

The ECP complex 134 responds to receipt of the DISCONNECT message by sending an MSC_NETWORK_RELEASE message via the cell controller 241 and multiple controller 244 to the channel element 245 handling the call. The channel element 245 responds by transmitting a RELEASE character to the mobile telephone 203 involved in the call and causing an FS_REMOVE packet 351 to be sent to the service line 612 handling the call. The FS_REMOVE character is the same as previously described for the mobile unit initiated disconnection and causes the same response.

In response to receiving the RELEASE signal, the mobile telephone 203 places the connection on-hook and transmits a HANGUP signal. This signal is received by the channel element 245 handling the connection and it responds by causing a RELEASE_CONFIRMATION message to be transmitted from the controller 241 of its cell to the ECP complex 134 to inform it of the shutdown of this side of the connection.

The ECP complex 134 responds by sending a CLEAR message to the DCS controller 261 of the speech encoder module 220 that previously handled the connection. The CLEAR message is the same as that described for the mobile unit initiated termination and causes the same response.

Figures 27-29 show the control signaling for soft handoff of the call from one cell 202 to another. Figure 27 shows the signaling for the start of soft handoff when a second cell 202, referred to as the follower cell, begins to handle the call together with the cell 202, referred to as the lead cell, which had been handling the call until then. A mobile telephone 203 participating in a call monitors the strength of pilot channel signals that it receives from a plurality of cells 202 including the lead cell 202 and sends the Control cell 202 periodically sends a report PWR.INFO. about these received power levels. The channel element 245 handling the connection forwards this report to the controller 241 of control cell 202. Based on this information and information exchanged between the cells 202 themselves, the controller 241 of control cell 202 determines whether only control cell 202 should continue to handle the connection or whether another cell 202 should be added to the connection. If the controller 241 of the master cell 202 determines that another cell 202 should be added to the connection and that this follower cell 202 can handle the connection using CDMA and the same mobile channel as the master cell 202, the controller 241 of the master cell 202 sends a HANDOFF_REQ message over control links 108 and IMS 104 to the controller 241 of the follower cell 202. The HANDOFF_REQ message conveys the DLCI of the service line 612 handling the connection that is not being used by the master cell 202 for this connection and the identification of the mobile channel on which the connection is being carried.

The controller 241 of the follower cell 202 receives the HANDOFF_REQ message and selects a channel element 245 of the follower cell 202 and one of the received DLCIs of the call-handling line 612 to process the call. (Alternatively, the HANDOFF_REQ message may convey the DLCI of the call-handling service line 612 used by the master cell 202 for that call, and the controller 241 of the follower cell 202 simply toggles the value of the least significant bit of that DLCI contained in the message to change the DLCI value to a second DLCI corresponding to the call-handling service line 612.) The controller 241 then forwards the selected DLCI, along with other contents of the received message, to the selected channel element 245 via a multi-controller 244. The selected channel element 245 prepares to process the call on the specified mobile channel and then causes the transmission of a FS_JOIN packet 351 to the service line 612 handling the connection. This packet uses the DLCI of the service line 612 received from the selected channel element 245 by the controller 241 as the packet address in field 302 and transmits the DLCI of the selected channel element 245 in its data field 304.

When the processor 602 serving the service line 612 handling the call receives the FS_JOIN packet, it sends an FS_ACK 351 packet back to the selected channel element 245 as an acknowledgement of receipt of the FS_JOIN packet as part of the LAPD processing in step 904 of Figure 11. The processor 602 then stores the transmitted DLCI of the selected channel element 245 as part of the connection state associated with the service line 612 handling the call and marks the connection state as being in the soft handoff state. A connection is now established between the selected channel element 245 of the follower cell 202 and the service line 612 handling the call and they begin exchanging call traffic packets.

The channel element 245 of the follow-up cell 202 reacts to the receipt of the FS_ACK packet by causing the transmission of a HANDOFF_ACK message by its cell controller 241 via control links 108 and IMS 104 to the controller 241 of the master cell 202 to inform it that the connection has been established. The controller 241 of the follow-up cell 202 also sends a HANDOFF_INFORMATION message to the ECP complex 134 to inform it of the soft handoff and the ECP complex 134 updates its database. Call traffic packets now flow between the one service line 612 and the channel elements 245 of the two master and follow-up cells 202 that are processing the connection.

Figures 28 and 29 show the signaling of the end of soft handoff when one of the two cells 202 handling the connection stops. This will typically, but not necessarily, indicate the Control cell 202. This scenario is shown in Figure 28.

During soft handoff, leader and follower cells 202 receive PWR.INFO. reports on pilot channel power levels measured by mobile phone 203. Note that this PWR.INFO. is different from the power control trend information received from both cells 202 by processor 602 and exchanged between the two cells 202 during soft handoff. Each cell 202 includes the received PWR.INFO. as a return signal in the next packet 350 that it sends to the service line 612 handling the call.

The processor 602 serving the service line 612 handling the call receives the PWR.INFO. as a return signal from both cells 202, selects and saves the PWR.INFO. from only one cell 202 in steps 968 of Figure 13 or 998 of Figure 14, and then sends the stored PWR.INFO. back to both cells 202 in steps 1216 and 1236 of Figure 15. Due to the actions taken by the processor 602, each cell 202 involved in the handoff receives PWR.INFO. sent by the cell 202 that received better quality signals from the mobile phone 203. The received PWR.INFO. is passed on to the controllers 241 of the receiving cells.

The controllers 241 use this information to determine when one of them should stop processing the call. When controller 241 of master cell 202 determines that it should stop processing the call, it sends a HANDOFF_DIRECTION signaling packet to processor 602 serving the call processing service line 612. The packet indicates that processing of the call is being handed over to slave cell 202. Processor 602 copies the signaling and sends it back to both master and slave cells 202, as shown in Figure 15.

Upon receipt of the HANDOFF_DIRECTION signal the channel elements 245 of both the master and slave cells 202 transmit the HANDOFF_DIRECTION information to the mobile telephone 203 for its information. The controller 241 of the master cell 202 then sends a MASTER_TRANSFER message over control links 108 and IMS 104 to the controller 241 of the other cell 202 participating in the soft handoff to notify it of the completion of the handoff and that it is to be the new master cell 202, and also passes a copy of this information to the channel element 245 of its own call-handling cell 202. The channel element 245 responds by not transmitting any further call traffic to and from the mobile telephone 203 and causing an FS_REMOVE packet to be sent to the call-handling line 612 to notify it of the termination of its participation in the call.

The processor 602 responds to the FS_REMOVE packet by sending the sending channel element 245 a FS_ACK packet as part of the protocol processing of the FS_REMOVE packet and by updating the connection state for the service line 612 to reflect that the connection is no longer in the soft handoff state. The controller 241 of the former master cell 202 receives the FS_ACK packet and responds by terminating its cell's participation in the connection. No further traffic for the connection flows between the channel element 245 of the former control cell 202 and the service line 612 handling the connection, but continues to flow between the service line 612 and the channel element 245 of the former follow-up cell 202. The controller 241 of the former control cell 202 now sends a HANDOFF_INFORMATION message to the ECP complex 134 to inform it of the completion of the handoff and its result. The ECP complex 134 updates its database accordingly.

It should be noted that the DCS controller 261 of the serving DCS 201 follows the procedures of Figures 27 and 28 remains completely uninvolved and that the ECP complex 134 is also not involved except to be notified of the completions of the procedures. As a result, the performance of the DCS controller 261 and the ECP complex 134 is not affected by the soft handoff procedures.

Figure 29 shows the scenario for the completion of soft handoff, with the follower cell 202 no longer serving the call and the master cell 202 continuing to serve the call alone. The procedure again begins with the master and follower cells 202 transmitting pilot channel reports PWR.INFO. to the processor 602 serving the call, and returning to both cells 202 the PWR.INFO. provided by the cell 202 receiving better signals from the mobile phone 203. If the controller 241 of the master cell 202 determines, based on these and other reports, that the follower cell 202 should terminate call processing, it sends a HANDOFF_DIRECTION signaling packet to the processor 602 indicating that the call processing is being taken over again by the master cell 202. The processor 602 copies the signaling and, as again shown in Figure 15, sends it back to both the lead cell and the follower cell 202.

Upon receipt of the HANDOFF_DIRECTION signal, the channel elements 245 of both the master and slave cells 202 transmit the HANDOFF_DIRECTION information to the mobile telephone 203 to inform it. The controller 241 of the master cell 202 then sends an INTRA/INTER_CELL HANDOFF_REMOVE message via control links 108 and IMS 104 to the controller 241 of the slave cell 202 to notify it that the handoff is complete and that it should exit the call. The controller 241 of the slave cell 202 notifies the channel element 245 of the slave cell 202 handling the call. The channel element 245 responds in the same manner as described in connection with Figure 28 for the channel element 245 of the master cell 202. described above: with the termination of the transmission of call traffic to and from the mobile telephone 203 and initiation of a packet exchange FS_REMOVE, FS_ACK with the processor 602. Traffic no longer flows between the channel element 245 of the follow-up cell 202 and the service line 612 handling the call, but continues to flow between the service line 612 and the channel element 245 of the master cell 202. The controller 241 of the former follow-up cell 202 now sends an INTRA/INTER_CELL_HANDOFF_ACK message to the master cell 202 and a HANDOFF_INFORMATION message to the ECP complex 134 to notify them of the completion of the handoff and its result. The ECP complex 134 updates its database accordingly.

As with Figure 28, there is little or no involvement of the DCS controller 261 and the ECP complex 134 in this handoff completion procedure.

Figure 30 shows the control signaling for call disconnection initiated by mobile telephone 203 while the call is in the soft handoff state. Mobile telephone 203 initiates call disconnection by transmitting a RELEASE character. This character is received by channel elements 245 handling the call in both the master cell and the slave cell 202. Each channel element 245 responds by sending a cell-to-mobile return signaling with the RELEASE character in the next packet 350 it sends to the service line 612 handling the call.

The processor 602 serving this service 612 receives the signaling from both cells 202, but saves only one copy in step 968 of Figure 13 or 998 of Figure 14 and sends the saved copy of the RELEASE signaling back to the channel elements 245 of both the master and slave cells 202 in the next traffic packet in steps 1216 or 1222 and 1236 of Figure 15. The controller 241 of the master cell 202 responds to the return of the RELEASE signaling by sending cell-mobile forward signaling MOBIL_DISCONNECT in the next packet 350 sent to the service line 612 handling the connection.

The processor 602 serving this service line 612 receives and stores the signaling at step 956 of Figure 13 and step 986 of Figure 14, respectively, and then sends it back to the channel elements 245 of both the master and slave cells 202 in the next traffic packet at steps 1222 and 1236 of Figure 15. Channel elements 245 of both the master and slave cells 202 each respond to receiving the MOBILE_DISCONNECT signaling by transmitting a RELEASE signal to the mobile telephone 203. The controller 241 of the master cell 202 then sends a NULL_TRAFFIC command in the cell-to-mobile signaling to the service line 612 in the next packet. This command is sent back by the processor 602 to both cells 202 in the manner just described for the MOBILE_DISCONNECT signaling. The channel elements 245 of both the master and slave cells 202 each respond to receipt of the NULL_TRAFFIC command by not transmitting any more call traffic and instead beginning to transmit null traffic to the mobile telephone 203. The two channel elements 245 also each cause an FS_REMOVE packet 351 to be sent to the service line 612 handling the call. The packets are the same as previously described and cause the same responses from the processor 602. Upon receipt of an FS_ACK packet from the processor 602, traffic to the mobile telephone 203 is stopped by the channel element 245 of each cell and causes a RELEASE_MSC message to be sent from its cell controller 241 to the ECP complex 134 to notify the complex 134 that the corresponding cell 202 is no longer handling the call. The ECP complex 134 updates its database accordingly and sends MSC_RELEASE_ACK messages to the controllers 241 of the master and slave cells 202. After receiving the second RELEASE_MSC message, the ECP complex 134 also sends a CLEAR message to the DCS controller 261 of the speech encoder module 220 processing the connection. The Message is the same as described for Figure 25 and causes the same response from the DCS controller 261.

Figure 31 shows the control signaling for call disconnection initiated by the public switched telephone network 100 while the call is in the soft handoff state. The network 100 releases the line 106 carrying the call. The release is detected by the controller 231 of the speech encoder module 220 handling the call and notified to the DCS controller 261, which in turn notifies the ECP complex 134 by sending it a DISCONNECT message.

The ECP complex 134 responds by sending an MSC_NETWORK_RELEASE message to cell controllers 241 of the lead and follower cells 202. The controller 241 of the lead cell 202 responds by sending cell-to-mobile unit forward signaling with a RELEASE character in the next packet 350 sent by it to the service line 612 handling the call.

The processor 602 serving this service line 612 receives the RELEASE character and stores it in step 956 of Figure 13 or step 986 of Figure 14 and then transmits the stored RELEASE signal to channel elements 245 of both the lead and follower cells 202 in the next traffic packet in steps 1222 and 1236 of Figure 15. Channel elements 245 of both the lead and follower cells 202 each respond to the signaling information by transmitting a RELEASE character to the mobile telephone 203 involved in the call.

In response to receiving the RELEASE characters transmitted by channel elements 245, the connection is placed on-hook by mobile telephone 203 and a MOBILE DISCONNECT character is transmitted as an acknowledgement. This character is received by channel elements 245 of both the master and slave cells 202. Each channel element 245 handling the connection responds by causing an FS_REMOVE packet 351 to be sent to the service line 612 handling the connection. The packets are the same as previously described and cause the same responses at processor 602. Upon receipt of the FS_ACK packet from processor 602, each channel element 245 responds by causing a RELEASE_CONFIRMATION message to be sent to ECP complex 134 to inform it of the disconnection of the connection.

After receiving the second message RELEASE_CONFIRMATION, the ECP complex 134 sends a CLEAR message to the DCS controller 261 of the speech encoder module 220 handling the connection. The message is the same as already described for Figure 25 and causes the same reaction.

Figure 32 shows the control signaling for a semi-soft handoff of the call from one channel element 245 to another. A semi-soft handoff occurs between channel elements 245 of either the same cell 202 or different cells 202 connected to the same DCS 201 and signifies a change in the mobile channel carrying the call. As with soft handoff, the controller 241 of the cell 202 handling the call - the serving cell - monitors the PWR.INFO. provided by the mobile telephone 203 to determine whether the serving channel element 245 should proceed with it or whether the call should be handed off to a new channel element 245 in either the same or another - new - cell 202. If the controller 241 of the serving cell 202 determines that it should handoff the call to a new channel element 245 and this new cell 202 can handle the call using CDMA, the controller 241 of the serving cell 202 sends a HANDOFF_REQ message over control links 108 and IMS 104 to the controller 241 of the new cell 202. (If the serving cell 202 and the new cell 202 are the same cell, this message is not sent outside the cell.) The message is the same as that described for soft handoff and causes the new cell 202 to perform the same Response as from a follower cell 202. However, since the new channel element 245 does not operate on the same mobile channel as the mobile telephone 203 and the serving channel element 245, the new channel element 245 is not in communication with the mobile telephone 203 and only null traffic packets flow from the new channel element 245 to the service circuit 612 handling the call.

The HANDOFF_ACK message returned from the new cell 202 to the serving cell 202 indicates the mobile channel on which the new channel element 245 is operating. The controller 241 of the serving cell 202 receives the HANDOFF_ACK message and responds by causing the serving channel element 245 to transmit a signal to the mobile telephone 203 telling it to switch its functions to the mobile channel on which the new channel element 245 is operating. When the mobile telephone 203 does so, traffic begins to flow between the mobile telephone 203, the new channel element 245, and the service line 612, but no longer flows between the mobile telephone 203 and the serving channel element 245, and only null traffic packets now flow from the serving channel element 245 to the service line 612.

The new channel element 245 responds to the start of receiving call traffic from the mobile telephone 203 by causing a HANDOFF_INFORMATION message to be sent to the ECP complex 134 and an INTERCELL_HANDOFF message to be sent to the serving cell 202 to inform it of the handoff. The ECP complex 134 updates its database while the controller 241 of the serving cell 202 causes the cell to drop out of servicing the connection. In particular, the channel element 245 of the serving cell 202 causes a FS_REMOVE packet to be sent to the service line 612 serving the connection. The packet is the same as previously described and causes the same response. Therefore, no further traffic flows between the serving channel element 245 and the service line 612. The serving channel element 245 responds to receipt of the FS_ACK packet from the service line 612 by causing a HANDOFF_INFORMATION message to be sent to the ECP complex 134 to notify it of the completion of the handoff, and the ECP complex 134 updates its database.

Again, it should be noted that the DCS controller 261 of the serving DCS 201 remains completely uninvolved in the process of Figure 31 and that the ECP complex 138 is also uninvolved, other than being notified of the completion of the process. As a result, the performance of the controller 261 and the ECP complex 134 is not affected by the semi-soft handoff process.

Figure 33 illustrates the control signaling for a hard handoff from one CDMA cell 202 to another. In CDMA, the hard handoff does not necessarily involve a change in the mobile channel, but it does involve a change in the cellular digital switch 201 (see Figure 2) that handles the call.

As with soft and semi-soft handoffs, the controller 241 of the cell 202 handling the call - referred to as the serving cell 202 - monitors the PWR.INFO. provided by the mobile telephone 203 and uses it along with other state information to determine whether the serving cell 202 should continue to handle the call or whether it should handoff the call to another cell 202, referred to as the new cell 202, that is connected to a different mobile telephone exchange 201 than the serving cell 202. When the controller 241 of the serving cell 202 determines that the call should be handed off, it sends a HARD_HANDOFF_REQUEST message to the ECP complex 134. The message identifies the call, the proposed new cell 202, and the mobile channel used by the serving cell 202 for the call.

The ECP complex 134 responds to the message by determining which DCS 201 is associated with the new cell 202, and within that DCS 201 selects a new speech encoder module 220 and a service line 612 of the new module 220 to handle the connection. The ECP complex 134 then selects a line 206 connecting the serving speech encoder module 220 of the serving DCS 201 to the new speech encoder module 220 of the new DCS 201 and sends the controller 261 of the serving DCS 201 a SETUP message identifying the selected new speech encoder module 220, the service line 612 and the line 206 and also identifying the line 106 of the serving speech encoder module 220 carrying the connection.

The controller 261 of the serving DCS 201 receives the SETUP message and responds by causing the controller 231 of the serving module 220 to seize the designated line 206 to output thereon the identification of the selected module 220 and the service line 612 and to connect the connection-carrying line 106 to the line 206 in a conference arrangement. This results in the seizure of the line 206 at the new module 220 and collection of the output identification by the controller 231 of the new module. The controller 261 of the serving DCS 201 then sends a CONNACK message to the ECP complex 134 to inform it of the establishment of the connection between the serving and the new module 220, while the controller 231 of the new module 220 sends the collected output information to the controller 261 of the new DCS 201, which sends an INCALL message with the collected output information to the ECP complex 134 to notify it of the incoming call.

The ECP complex 134 associates the received CONNACK and INCALL messages based on their content; the messages serve as confirmation to the ECP complex 134 that the TDM buses 130 of the new and serving modules 220 are now interconnected via line 206. The ECP complex 134 then sends a message MSC_NEW_HANDOFF to the controller 241 of the new cell 202. This message notifies the new cell 202 that it has been selected to handle the call and conveys to it the identification of the mobile channel currently carrying the call. The controller 241 of the new cell responds by determining whether the new cell 202 can handle the call and, if so, on which mobile channel. The controller 241 of the new cell then sends a CHANNEL_ACTIVATION_CONFIRMATION message with this information back to the ECP complex 134. Assuming the new cell 202 can handle the call, the ECP complex 134 sends the controller 241 of the new cell an MSC_FS_ASSIGNMENT message with the DLCI of the service line 612 of the new module 220 that has been selected to handle the call. This message is the same as that discussed in connection with Figure 23 and causes the same responses. The new cell 202 sends a FS_CONFIRMATION message back to the ECP complex 134 and the ECP complex 134 in turn sends a MSC_OLD_HANDOFF message to the serving cell 202 informing it of the establishment of the connection between the new channel element 245 and the new service line 612 and the mobile channel on which the new channel element 245 operates.

The ECP complex 134 responds to the FS_CONFIRMATION message by sending an ACCEPT message to the controller 261 of the new DCS 201. The controller 261 of the new DCS 201 responds in the manner already described for ACCEPT messages by causing the controller 231 of the new module 220 to establish a connection between the new service line 612 and the line 206 connecting the new module 220 to the serving module 220. As a result, the output of both the new and serving service lines 612 are connected in a conference arrangement to the same time slot of the TDM bus 130 of the serving speech encoder module 220. If both the new and serving channel elements 245 operate on the same mobile channel, this results in superposition of the duplicated outputs in the same time slot and therefore substantially no influence on the content of the time slot. If the two channel elements 245 are not operating on the same mobile channel, this results in superposition of samples of real traffic and zero traffic - voice or data or silence - in the same time slot and therefore again substantially no influence on the content of the time slot. The controller 261 of the new DCS 201 then sends a CONNACK message to the ECP complex 134 to notify it of the establishment of the connection. The controller 231 of the serving module 220 detects the establishment of the connection and notifies the controller 261 of the serving DCS 201, which sends an ANSWER message back to the ECP complex 134 to inform it of this.

The serving cell controller 241 responds to the MSC_OLD_HANDOFF message received from the ECP complex 134 by examining the message content to determine whether the new channel element 245 is operating on the same mobile channel as the serving channel element 245. If not, the serving cell controller 241 causes the serving channel element 245 to transmit a signal to the mobile telephone 203 commanding it to switch operation from the mobile channel it is currently using to the mobile channel used by the new channel element 245, as shown in dashed lines in Figure 33. If the mobile telephone 203 does so, the traffic flow is switched from the serving cell 202 to the new cell 202, as shown in dashed lines.

The channel element 245 of the new cell 207 responds to the start of reception of the call traffic by causing the controller 241 of the new cell to send a HANDOFF_VOICE_CHANNEL_CONFIRMATION message to the ECP complex 134. This message informs the ECP complex 134 of the successful handoff. The ECP complex 134 responds by sending a MSC_CHANNEL_DEACTIVATION message to the serving cell 202 and a CLEAR message to the controller 261 of the serving DCS 201 to cause the serving cell 202 and the serving SPU 264 to exit from processing the call.

The controller 241 of the serving cell 202 forwards the MSC_CHANNEL_DEACTIVATION message to the serving channel element 245, which responds by causing a FS_REMOVE packet to be forwarded to the serving service line 612. The packet is the same as previously described and causes the same response. When the serving cell 202 is no longer handling the connection, its controller 241 sends a FS_CONFIRMATION message to the ECP complex 134 to inform it of this.

The controller 261 of the serving DCS 201 forwards the received CLEAR message to the controller 231 of the serving module 220. The controller 231 responds by causing the conversion and maintenance processor 609 of the speech processing unit 264 containing the serving service line 612 to switch the connection (i.e., the time slot of the TDM bus 130 carrying the connection) from the time slot assigned to that service line 612 on the collective bus 607. However, since the new service line 612 of the new module 220 is now connected to the line 106 carrying the connection over line 206 to and from the TDM bus 130 of the serving module 220, the controller 231 of the serving module 220 does not release this line 106 and the timing of the TDM bus 130. The controller 261 of the serving DCS 201 then sends a CLEAR_ACK message to the ECP complex 134 to notify it that the serving SPU 264 of the serving module 220 is no longer serving the connection. Receiving both the CLEAR_ACK and FS_CONFIRMATION messages indicates to the ECP complex 134 that the handoff is complete.

Figures 34-35 illustrate the control signaling for a hard handoff from a CDMA radio 243 of a serving cell 202 to a regular analog radio 143 of a new cell 102 or 202. Figure Figure 34 shows the control signaling for handover between two cells connected to the same DCS 201 and Figure 35 shows handover between two cells connected to different DCS 201.

Referring to Figure 34, a conventional mobile telephony cell 102 may be equipped with a CDMA pilot channel. If so, then the control communications with a new cell 102 are as with a new cell 202 and are shown in Figure 33; if the new cell 102 is not equipped with a CDMA pilot channel, then the control communications shown in Figure 34 for the new cell 102 are also with the serving cell 202 instead. In other words, if the new cell 102 is not equipped with a CDMA pilot channel, then the conversion of the call to conventional mobile telephony occurs at the serving cell 202 and the call is only thereafter handed off from the serving cell 202 to the new cell 102 in the conventional hard handoff manner.

As with handoff types previously discussed, the controller 241 of the serving cell 202 checks the PWR.INFO. provided by the mobile telephone 203 to determine whether or not to handoff the call to another cell. If the controller 241 of the serving cell 202 determines that it should handoff the call to a conventional radio 143 in a cell 202 or 102 and the new cell 202 or 102 is connected to the same mobile telephone exchange 201 as the serving cell 202, the controller 241 sends an ANALOG_HANDOFF_REQUEST message to the ECP complex 134. The message identifies the proposed new cell 102 or 202. The ECP complex 134 responds by selecting a line 109 of a switching module 120 or 220 to which the new cell 102 or 202 is connected and sending a message MSC_NEW_HANDOFF to the controller 141 or 241 of the new cell 102 or 202. The selected line 109 is identified in the message and is requested, whether the new cell 102 or 202 can handle the connection. The controller 141 or 241 of the new cell 102 or 202 responds with a CHANNEL_ACTIVATION_CONFIRMATION message to the ECP complex 134 identifying the conventional mobile channel on which it will handle the connection and also connects that mobile channel to the selected line 109. The ECP complex 134 responds by selecting a line 109 connected to the serving module 220 and sending a CONNECT message to the DCS controller 261 of the serving DCS 201 identifying the new module 120 or 220 to which the new cell 102 or 202 is connected, the selected line 109 connected to the new module 120 or 220, and the selected line 109 outgoing from the serving speech encoder module 220.

The DCS controller 261 of the serving DCS 201 receives the CONNECT message and responds by causing the controller 231 of the serving module 220 to connect the connection (the time slot on the TDM bus 130) in a conference arrangement to the designated outgoing line 109 and causing the TMS 121 to connect the two designated lines 109 together. The controller 261 of the serving DCS 201 then sends a CONNACK message to the ECP complex 134 to inform it of the establishment of the connection between the serving and new modules.

The ECP complex 134 responds by sending an MSC_OLD_HANDOFF message to the controller 241 of the serving cell 202 conveying the mobile channel on which the new cell 102 or 202 will handle the call. In response, the controller 241 causes the serving channel element 245 to transmit a signal to the mobile telephone 203 instructing it to switch to conventional mobile telephone operation and to use the mobile channel specified in the MSC_NEW_HANDOFF message.

If the mobile phone 203 does this and with When transmission on the new mobile channel begins, the new cell 102 or 202 receives transmissions and notifies the ECP complex 134 via a HANDOFF_VOICE_CHANNEL_CONFIRMATION message. The ECP complex 134 responds with an MSC_CHANNEL_DEACTIVATION message to the serving cell 202 and a CLEAR message to the DCS controller 261 of the serving DCS 201 to cause the serving cell 202 and serving SPU 264 to exit handling the call. The messages are the same as those discussed for CDMA-CDMA hard handoff and result in the same responses. As in this case, receipt of both CLEAR_ACK and FS_CONFIRMATION messages indicates to the ECP complex 134 that the handoff is complete.

Referring now to Figure 35, handoff to a new cell 102 or 202 connected to a different switch 101 or 201 than the serving cell 202 begins in the same manner as shown in Figure 34. However, after a decision to handoff to a cell 102 served by a new DCS 101 or 201, the controller 241 of the serving cell 202 sends an ANALOG_HANDOFF_REQUEST message to the ECP complex 134 to request the handoff. The message identifies the proposed new cell 102 or 202. The ECP complex 134 responds to this message by determining which switch 101 or 201 is connected to the new cell 102 or 202, respectively, and selecting a new switch module 120 or 220 of that switch 101 or 201, respectively, and a line 106 connected to that selected module 120 or 220, respectively, to handle the connection. Thereafter, the ECP complex 134 selects an outgoing line 106 connected to the serving module 220 and sends a SETUP message to the DCS controller 261 of the serving DCS 201 identifying the selected new module 120 or 220 and its connected line 106, the line 106 outgoing from the serving speech encoder module 220, and the line 106 of the serving speech encoder module 220 carrying the connection.

The SETUP message is analogous to that described in connection with Figure 33 and causes the same reactions. The handover therefore proceeds as described for Figure 33. However, no SPU 264 will be involved in processing the connection at the new DCS 101 or 201 and the ECP complex 134 therefore proceeds directly to sending an ACCEPT message to the DCS controller 161 or 261 of the new DCS 101 or 201, respectively, instead of sending an FS_ASSIGN message to the new cell 102 or 202, respectively, as in Figure 33. The DCS controller 161 or 261 responds by causing the controller 131 of the new module 120 or the controller 251 of a cell interconnect module 209 to connect the selected line 106 of the new module 120 or 220 to the connection (i.e., the corresponding connection timing on either the TDM bus 130 of the module 120 or the TDM bus 230 of the dM 209), thereby establishing a connection between that selected line 106 and the new cell 102 or 202. As in Figure 33, this results in the output of both the new cell 102 or 202 and the serving cell 202 being connected to the same timing on the TDM bus 130 of the serving speech encoder module 220. The DCS controller 161 or 261 of the new DCS 101 or 201 then sends a CONNACK message back to the ECP complex 134 to inform it that the connection has been established, while the controller 231 of the serving module 220 detects that the connection has been established and notifies the controller 261 of the serving DCS, which responds by sending an ANSWER message back to the ECP complex 134.

The ECP complex 134 responds to receipt of the CONNACK message by sending a message MSC_OLD_HANDOFF to the controller 241 of the serving cell 202. The message is the same as already discussed in connection with Figure 34 and the handoff proceeds from now on exactly as described for Figure 34 until the handoff is completed.

The specialist will of course be given various Changes and modifications to the embodiment described above are apparent. For example, different packet transmission methods such as Asynchronous Transfer Mode (ATM) may be used. Or the division of functionality between the control entities of the cells, the ECP complex and the cellular digital switches may be changed. Also, modules within a cellular digital switch (both CIM 209 and SCM 220) may be interconnected by a central switching stage rather than simply directly via lines. Furthermore, the system described above may be applied to pseudosynchronous wireless access systems other than mobile telephony - for example, to personal communications networks (PCNs).

Claims (18)

1. Communication system with the following: a first unit for at least one of (a) sending outgoing communication traffic and (b) receiving incoming communication traffic at times dictated by first clock signals having a nominal frequency and a first phase; a second unit for transmitting at least one of the received (a) incoming communications traffic and (b) outgoing communications traffic at times dictated by second clock signals having the nominal frequency; a third unit for interconnecting communications between the first and second units by at least one of (a) sending to the second unit outgoing communication traffic received from the first unit at times dictated by third clock signals having the nominal frequency and having a second phase offset from the first phase by an adjustable fixed first amount, and (b) sending to the first unit incoming communication traffic received from the second unit at times dictated by fourth clock signals having the nominal frequency and having a third phase offset from the first phase by an adjustable fixed second amount; a communication medium connecting the second unit to the third unit and carrying communication traffic transmitted by the second unit or the third unit to the third unit or the second unit for reception, the communication medium having a variable transmission delay; Runtime determination means with at least one of (a) first means for determining whether the second unit receives outgoing communication traffic from the third unit within first predetermined time windows prior to the times of sending the received outgoing receives communication traffic through the second unit, and (b) second means for determining whether the third unit receives incoming communication traffic from the second unit within second predetermined time windows prior to times of transmission by the third unit of the received incoming communication traffic; and third means responsive to a determination that either receptions of the outgoing communication traffic at the second unit fall outside the first windows or receptions of the incoming communication traffic at the third unit fall outside the second windows for either increasing or decreasing the amount of offset of either the second phase or the third phase from the first phase to shift the receptions falling outside their respective windows into the respective windows. 2. The system of claim 1, wherein the third means is responsive to a determination by the first means that receptions of the outgoing communication traffic at the second unit lag the first windows to reduce the amount of offset of the second phase from the first phase. 3. The system of claim 2, wherein the third means is further responsive to a determination by the first means that receptions of the outgoing communication traffic at the second unit are leading the first windows to increase the amount of offset of the second phase from the first phase. 4. The system of claim 1, wherein the third means is responsive to a determination by the second means that receptions of the incoming communication traffic at the third unit lag the second windows for increasing the amount of offset of the third phase from the first phase. 5. The system of claim 4, wherein the third means is further responsive to a determination by the second means that receptions of the incoming communication traffic on the third unit, precede the second windows to reduce the amount of offset of the third phase from the first phase. 6. The system of claim 1, wherein the third means adjusts the first and second amounts of offset at the start of a communication in one step each by any amount necessary to shift the receptions falling outside their respective windows into the respective windows, and adjusts the first and second amounts of offset during the communication in a series of sequential steps by an integer multiple of a same predetermined amount during each step. 7. The system of claim 1, further comprising: fourth means, cooperating with the third means, for inserting additional traffic into the outgoing communication traffic sent by the first unit and causing the additional traffic to be received by the third unit while increasing the amount of the second phase offset, and deleting a portion of the outgoing communication traffic transmitted by the first unit from the outgoing communication traffic received by the third unit while decreasing the amount of the second phase offset, and for inserting additional traffic into the incoming communication traffic received by the third unit and sending the additional traffic to the first unit while increasing the amount of the third phase offset, and deleting a portion of the incoming communication traffic received by the third unit from the incoming communication traffic transmitted to the first unit while decreasing the amount of the third phase offset. 8. The system of claim 1, wherein: the first unit is designed to send a stream of outgoing communication traffic and to receive a stream of incoming communication traffic; the second unit is designed to send packets of received incoming communication traffic and to receive packets of outgoing communication traffic for transmitting the outgoing communication traffic; the third unit contains: fourth means, responsive to receipt of the stream of outgoing communication traffic from the first unit, for packetizing the received outgoing communication traffic and sending the packets of the received outgoing communication traffic to the second unit at times dictated by the third clock signals, and fifth means, responsive to receipt of the packets of incoming communication traffic from the second unit, for depacketizing the received incoming communication traffic and sending the depacketized received incoming communication traffic to the first unit at times dictated by the fourth clock signals; the first means for determining whether the second unit receives the packets of the outgoing communication traffic from the third unit within the first predetermined windows; and the second means for determining whether the third unit receives the packets of incoming communication traffic from the second unit within the second predetermined windows. 9. A call traffic processing device for a cellular radiotelephone system, comprising the device, a communication medium connected to the device having a varying transmission delay, at least one cell connected to the communication medium, and each for transmitting first packets containing first frames of coded call traffic received from a radiotelephone device arriving at the device via the medium, and for transmitting outgoing call traffic received from the device via the medium to the radiotelephone device in second packets containing second frames of coded outgoing call traffic and a mobile telephone switching system for interconnecting cells with each other and with a telephone network by routing a first digital stream of incoming call traffic received by the device to a destination and by routing a second digital stream of outgoing call traffic received from a source to the device, and wherein the first and second digital streams are synchronized with first clock signals having a nominal frequency and a first phase derived from the telephone network and the transmissions of the incoming and outgoing call traffic through the cell are synchronized with second clock signals having the nominal frequency, the call traffic processing means comprising: vocoder means for receiving the second digital stream of outgoing call traffic synchronously with the first clock signals, encoding the received outgoing call traffic and transmitting second frames of the encoded outgoing call traffic synchronously with third clock signals having the nominal frequency and a fourth phase offset from the first phase by an adjustably fixed third amount; Vocoding means for receiving first frames of the encoded incoming call traffic synchronously with fourth clock signals having the nominal frequency and a third phase offset from the first phase by an adjustable fixed second amount, decoding the received incoming call traffic and transmitting the second digital stream of incoming call traffic synchronously with the first clock signals; processing means for receiving the second frames of encoded outgoing call traffic from the vocoder means, packetizing the received second frames into the second packets and transmitting the second packets to the cell synchronously with fifth clock signals having the nominal frequency and a fourth phase offset from the first phase by an adjustable fixed third amount; processing means for receiving the first packets from the cell, depacketizing the received first packets into the first frames and transmitting the first frames to the vocoder means synchronously with sixth clock signals having the nominal frequency and a fifth phase offset from the first phase by an adjustable fixed fourth amount; Clock signal generating means for deriving the third, fourth, fifth and sixth clock signals from the first clock signals; first means for determining whether the processing means incomingly receive the first packets within first predetermined time windows prior to the times of transmission of the first frames by the processing means incoming to the vocoder means; second means for determining whether the cell receives the second packets within second predetermined time windows prior to the times of transmission by the cell of the received outgoing call traffic to the mobile telephone device; and third means responsive to a determination that either receptions of the first packets fall outside the first windows or receptions of the second packets fall outside the second windows for causing the clock signal generating means to either increase or decrease the amounts of the offsets of either both the second and fourth phases or both the third and fifth phases with respect to the first phase to shift the packet receptions falling outside their respective windows into the respective windows. 10. A method of operating a communications system comprising a first unit for at least one of (a) sending outgoing communications traffic and (b) receiving incoming communications traffic, a second unit for transmitting at least one of the received (a) incoming communications traffic and (b) outgoing communications traffic, a third unit for interconnecting communications between the first and second units, and a communication medium with a fluctuating transmission time connecting the second unit to the third unit, the method comprising the following steps: at least one of (a) sending outgoing communications traffic from the first unit and (b) receiving incoming communications traffic at the first unit at times dictated by the first clock signals having a nominal frequency and a first phase; transmitting at least one of received (a) incoming communication traffic and (b) outgoing communication traffic from the second unit at times dictated by second clock signals having the nominal frequency; at least one of (a) receiving at the third unit the outgoing communication traffic transmitted by the first unit and transmitting from the third unit outgoing communication traffic received from the first unit to the second unit at times dictated by third clock signals having the nominal frequency and having a third phase offset from the first phase by an adjustable fixed first amount, and (b) receiving incoming communication traffic transmitted by the second unit at the third unit and transmitting incoming communication traffic received by the second unit from the third unit to the first unit at times dictated by fourth clock signals having the nominal frequency and having a third phase offset from the first phase by an adjustable fixed second amount; at least one of (a) transmitting the outgoing communication traffic transmitted by the third entity over the communication medium to the second entity for receipt, and (b) transmitting the incoming communication traffic transmitted by the second unit over the communication medium to the third unit for receipt; at least one of (a) determining whether the second unit receives outgoing communication traffic from the third unit within first predetermined time windows prior to the times of sending the received outgoing communication traffic by the second unit and (b) determining whether the third unit receives incoming communication traffic from the second unit within two predetermined time windows prior to the times of transmission of the received incoming communication traffic by the third unit; and either increasing or decreasing the amount of offset of either the second phase or the third phase from the first phase in response to a determination that either receptions of the outgoing communication traffic at the second unit fall outside the first windows or receptions of the incoming communication traffic at the third unit fall outside the second windows to shift the receptions that fall outside their respective windows into the respective windows. 11. The method of claim 10, wherein the step of either increasing or decreasing the amount of offset comprises the step of decreasing the amount of offset of the second phase from the first phase in response to a determination that receptions of the outgoing communication traffic at the second unit lag the first windows. 12. The method of claim 11, wherein the step of either increasing or decreasing the amount of offset further comprises the step of increasing the amount of offset of the second phase from the first phase in response to a determination that receptions of the communications traffic originating at the second unit are ahead of the first windows. 13. The method of claim 10, wherein the step of either increasing or decreasing the amount of offset comprises the step of increasing the amount of offset of the third phase from the first phase in response to a determination that receptions of the communication traffic arriving at the third unit lag the second windows. 14. The method of claim 13, wherein the step of either increasing or decreasing the amount of offset further comprises the step of decreasing the amount of offset of the fourth phase from the first phase in response to a determination that receptions of the communication traffic arriving at the third unit lead the second windows. 15. The method of claim 10, wherein the step of either increasing or decreasing the amount of offset comprises the steps of: adjusting the first amount and the second amount of offset at the start of a communication, each in one step, by any amount necessary to shift the receptions that fall outside their respective windows into the respective windows; and Adjusting the first amount and the second amount of offset during communication in a series of sequential steps by an integer multiple of a same predetermined amount during each step. 16. The method of claim 10, further comprising the following steps: inserting additional traffic into the outgoing communication traffic transmitted by the first unit and causing the third unit to receive the additional traffic while increasing the amount of the offset of the second phase; Deleting part of the data from the first unit transmitted outgoing communication traffic from the outgoing communication traffic received by the third unit, while reducing the amount of the second phase offset; inserting additional traffic into the incoming communication traffic received by the third unit and transmitting the additional traffic to the first unit while increasing the amount of the third phase offset; and Deleting a portion of the incoming communication traffic received by the third unit from the incoming communication traffic transmitted to the first unit while reducing the amount of the third phase offset. 17. Method according to claim 10 in a communication system, wherein the first unit is intended for transmitting a stream of outgoing communication traffic and receiving a stream of incoming communication traffic and the second unit is intended for transmitting packets of received incoming communication traffic and receiving packets of outgoing communication traffic for transmitting the outgoing communication traffic, wherein the step of receiving the outgoing communication traffic transmitted by the first means at the third unit comprises the following steps: Receiving the stream of outgoing communication traffic from the first unit and Packeting the received outgoing communication traffic; the step of transmitting outgoing communication traffic from the third unit to the second unit the step of transmitting the packets of the received outgoing communication traffic to the second unit at times dictated by the third clock signals; the step of receiving the communication traffic transmitted by the second unit and arriving at the third unit comprises the following steps: Receiving the packets of incoming communication traffic from the second unit and Depacketizing the received incoming communication traffic; the step of transmitting incoming communication traffic from the third unit to the first unit, the step of transmitting the depacketized received incoming communication traffic to the first unit at times dictated by the fourth clock signals; the step of determining whether the second unit receives outgoing communication traffic the step of Determining whether the second unit receives the packets of outgoing communication traffic from the third unit within the first predetermined window; and the step of determining whether the third unit receives incoming communication traffic, the step of Determining whether the third unit receives the packets of incoming communication traffic from the second unit within the second predetermined window. 18. A method of processing call traffic in an interface arrangement of a cellular radiotelephone system comprising the arrangement, a communications medium connected to the device having a varying transmission delay, at least one cell connected to the communications medium, each for transmitting first packets containing first frames of coded call traffic received from a radiotelephone device and arriving at the arrangement via the medium, and for transmitting call traffic received from the arrangement via the medium to the radiotelephone device in second packets containing second frames of coded outgoing call traffic, and a mobile telephone switching system for interconnecting cells with one another and with a telephone network by routing a first digital stream of incoming call traffic received by the arrangement to a destination and by routing a second digital stream of outgoing call traffic received from a source to the arrangement, and wherein the first and second digital streams are synchronized with first clock signals having a nominal frequency and a first phase derived from the telephone network and the transmissions of the incoming and outgoing call traffic through the cell are synchronized with second clock signals having the nominal frequency, the method comprising the steps of: Receiving the second digital stream of outgoing call traffic synchronously with the first clock signals; Encoding the received outgoing call traffic; transmitting second frames of the coded outgoing call traffic synchronously with third clock signals having the nominal frequency and a second phase offset from the first phase by an adjustable fixed first amount; Receiving the second frames of the coded outgoing call traffic; Packeting the received second frames into the second packets; transmitting the second packets to the cell over the communication medium synchronously with fifth clock signals having the nominal frequency and a fifth phase offset from the first phase by an adjustable fixed third amount; Receiving the first packets from the cell via the communication medium; Depacketizing the received first packets into the first frames; transmitting the first frames synchronously with sixth clock signals having the nominal frequency and a fifth phase offset from the first phase by an adjustable fixed fourth amount; Receiving the first frames of the coded incoming call traffic synchronously with fourth clock signals having the nominal frequency and a third phase which differs by an adjustable fixed second amount from the first phase is offset; Decoding the received incoming call traffic; transmitting the second digital stream of incoming call traffic synchronously with the first clock signals; determining whether the interface arrangement receives the first packets within first predetermined time windows before the transmission times of the first frames by the interface arrangement; determining whether the cell receives the second packets within two predetermined time windows prior to the transmission times by the cell of the received outgoing call traffic to the mobile telephone device; and either increasing or decreasing the amounts of the offsets of either both the second and fourth phases or both the third and fifth phases with respect to the first phase in response to a determination that either receptions of the first packets fall outside the first windows or receptions of the second packets fall outside the second windows, to shift the packet receptions that fall outside their respective windows into the respective windows.